Author: Husham Al-Ani Affiliation: Swiss International University (SIU) ORCID ID: 0009-0006-5269-9222 Doi: https://doi.org/10.65326/u7y10030 Submitted 08 May 2026; Revised 18 June 2026; Revised 08 July 2026; Revised 18 August 2026; Accepted 21 August 2026; Available online 28 August 2026; Version of Record 28 August 2026. Volume 3, December 2026, (10030) Abstract This paper presents a critical integrative review of how artificial intelligence-driven knowledge management may support decision-making quality in medium-sized technology companies. It argues that generative AI shifts the central knowledge management challenge from retrieval to trustworthiness. While generative AI improves access to dispersed organisational knowledge, its outputs may lack traceable sources, contain confident errors, or blur the boundary between reliable knowledge and probabilistic text. Drawing on classical knowledge management theory, recent AI and generative AI research, decision-making literature and regional implementation evidence, the paper develops a conceptual framework in which knowledge management practices mediate the relationship between AI-driven knowledge management and decision-making quality. Perceived challenges such as poor data quality, weak governance, limited skills and overreliance on AI may weaken this relationship. The paper identifies provenance, validation, governance and human review as core practices for trustworthy AI-supported decision-making. Keywords: artificial intelligence; generative AI; knowledge management; decision-making quality; medium-sized technology companies; integrative review; organisational knowledge. 1. Introduction Knowledge management has been discussed for more than three decades as a central issue in organisational work. Classical studies showed that knowledge is not only stored information. It includes experience, interpretation and the ability to use what is known in action (Davenport & Prusak, 1998). Nonaka and Takeuchi (1995) also showed that organisational knowledge is created through the movement between tacit and explicit knowledge. These ideas remain important because many companies still struggle to turn individual experience into knowledge that can be reused. The issue is more visible in technology companies. A single decision may require technical knowledge, supplier history, project experience, client requirements and financial judgement. Choo (1998) explained that organisations use information to construct meaning and guide decisions. This means that decision-making is not a mechanical process. It depends on organisational memory and on the ability to interpret knowledge in context. Medium-sized technology companies represent a useful context for this discussion. They often have rich practical knowledge gained from projects, suppliers and technical work. At the same time, their knowledge systems may be less formal than those of large corporations. Important files may be spread across emails, project folders, technical documents and personal memory. This creates a gap between what the company has learned and what it can use when a decision is needed. This context requires separate attention because medium-sized technology companies occupy a position between start-ups and large corporations. Unlike start-ups, they usually have accumulated project experience, supplier records, client histories and technical knowledge. Unlike large corporations, they may not have mature knowledge management systems, specialised data teams or formal governance structures. This creates a specific organisational condition in which valuable knowledge exists, but its use in decision-making may remain uneven. For this reason, AI-driven knowledge management may be especially relevant to such firms, provided that it is connected to clear routines for documentation, retrieval, sharing and human review. Artificial intelligence has created new possibilities for reducing this gap. It can support search, classification, summarisation and the comparison of previous cases. Yet the value of these tools should not be assumed. Davenport and Ronanki (2018) warned that organisations gain more from artificial intelligence when it is attached to clear business problems rather than treated as a general technology fashion. The same logic applies to knowledge management. An intelligent tool can only work well when the knowledge base and organisational routines are prepared. This review uses the term medium-sized technology companies as an organisational category referring to technology-related firms with approximately 50 to 250 employees. This numerical benchmark follows the European Commission definition of a medium-sized enterprise (European Commission, 2003), while the paper uses it as a practical guide rather than a universal legal threshold. The analytical focus is on firms that have accumulated project, supplier and client knowledge but lack the mature knowledge management infrastructure, specialised data teams or formal governance systems found in larger corporations. Definitions and thresholds vary across countries and sectors, particularly in the Arab region; accordingly, organisational complexity and the degree of knowledge-system formalisation remain more important here than headcount alone. The literature on this topic remains distributed across several related streams. Classical knowledge management studies provide the foundation for understanding knowledge creation, knowledge sharing and organisational memory. However, much of this work was developed before the current expansion of AI and before the rise of generative AI in organisational knowledge work. Recent studies on AI and knowledge management explain how intelligent tools may support search, classification, retrieval, summarisation and decision support. Yet many of these studies still discuss AI as a general digital capability or focus mainly on large organisations and broad digital transformation. Less attention is given to medium-sized technology companies, where practical knowledge is often rich but not always formally organised. The recent development of generative AI makes this gap more important. Earlier knowledge systems mainly helped organisations store and retrieve what was already documented. Generative AI can also produce summaries, combine previous material and create knowledge-like outputs. This does not remove the need for knowledge management. It increases the need for stronger knowledge practices. In AI-supported organisations, knowledge must not only be captured and shared. It must also be checked, traced to its source, governed and reviewed by people who understand the organisational context. Regional and emerging-economy implementation evidence adds contextual insight on readiness, skills, infrastructure and governance barriers — though it rarely explains how AI-driven knowledge management may influence decision-making quality through knowledge management practices specifically. The research gap addressed in this paper therefore lies in the limited integration of these streams. Existing literature does not yet provide a focused conceptual explanation of how AI-driven knowledge management, knowledge management practices, decision-making quality and perceived challenges work together in medium-sized technology companies. The gap is not only about whether AI supports knowledge work, but about how generative AI changes the conditions under which organisational knowledge can be trusted and used in decisions. The theoretical contribution of this paper rests on a specific claim about what generative AI changes in knowledge management. Before generative AI, retrieval was often experienced as the most visible bottleneck in medium-sized technology companies because employees struggled to find relevant knowledge dispersed across files, emails, project records and individual memory. Generative AI can reduce the salience of that bottleneck by making access faster and more conversational, but it does not eliminate the underlying problems of retrieval, grounding or source selection. Instead, it adds a verification burden: employees may obtain fast and fluent answers without knowing where those answers came from, whether they are accurate, or who is responsible if they are wrong. This paper therefore argues that generative AI shifts the principal organisational emphasis from retrieval alone to trustworthiness. Provenance, validation, governance and human review become central knowledge management practices alongside documentation and retrieval. This argument extends classical knowledge management theory and gives particular attention to medium-sized technology companies, where practical knowledge is often rich but informally organised and where ungoverned AI use may have consequential effects on decisions. What this paper adds to existing work Storey (2025) is an important recent contribution, proposing a framework for knowledge management in the generative AI era and examining how generative artificial intelligence (GenAI) affects novice and expert knowledge workers. The present paper complements that work by focusing on medium-sized technology companies and by specifying provenance, validation, governance and human review as the practices required to make AI-supported knowledge trustworthy. However, the mediation structure itself is not new. Leoni et al. (2022) empirically showed, in a sample of 120 senior executives from Italian manufacturing firms, that knowledge management processes mediate the effects of AI on supply-chain resilience and firm performance. The novelty claimed here therefore rests not on proposing mediation in the abstract, but on identifying how the content and priority of the mediator change under generative AI: from an emphasis on documentation and retrieval toward organisational routines for provenance, validation, governance and human review. Table 1 summarises this positioning. Table 1 How this paper extends Storey (2025) Dimension Storey (2025) This paper Organisational scope All organisations and knowledge workers Medium-sized technology companies specifically Central argument GenAI affects novice vs expert workers differently; risks include knowledge loss and over-automation GenAI shifts the central KM challenge from retrieval to trustworthiness Role of KM practices Contextual background; general framework Mediator whose distinctive content under GenAI is provenance, validation, governance and human review New KM practices identified Quality evaluation by expert workers Provenance, validation, governance and human review as core organisational practices Empirical testing path Not specified Six propositions for future empirical testing in medium-sized technology firms Closest mediation precedent Not the paper’s focus Leoni et al. (2022) tested KMP mediation for AI effects on manufacturing outcomes; this paper re-anchors novelty in the trustworthiness content of the mediator Before presenting the research problem, it is useful to clarify how the main concepts are used in this review. The definitions below are not intended as universal definitions. They are working definitions developed to support the logic of the present paper and the proposed conceptual framework. Table 2 presents the working definitions used throughout the review. Table 2 Working definitions of the main concepts Concept Working definition in this paper AI-driven knowledge management The organisational use of artificial intelligence tools to support the acquisition, classification, storage, retrieval, sharing and application of knowledge in a way that improves access to relevant organisational experience during work and decision-making. Knowledge management practices The organisational routines and procedures through which knowledge is identified, documented, validated, stored, shared, updated and applied in operational and managerial decision situations. Decision-making quality The extent to which organisational decisions are informed by relevant knowledge, made in a timely manner, consistent with organisational objectives and supported by evidence, experience and professional judgement. Perceived challenges The organisational and technical barriers that may weaken AI-supported knowledge work, including poor documentation, limited infrastructure, skill gaps, resistance and governance concerns. Critical integrative review A review approach that connects related bodies of literature, evaluates their contribution and uses them to develop a conceptual argument or framework. These definitions are used to connect the reviewed literature with the proposed conceptual framework. They also clarify that the paper treats artificial intelligence as part of a wider knowledge process rather than as an isolated technical tool. 2. Research Problem and Review Questions The issue addressed in this paper is not the weakness of the existing literature, but its fragmentation across several related research streams. Classical studies explain knowledge creation, knowledge sharing and knowledge systems. Recent studies explain the role of artificial intelligence and generative AI in knowledge work. Regional and emerging-economy implementation evidence adds contextual insight into readiness, skills, governance and implementation barriers. What is still missing is an integrated explanation of how artificial intelligence-driven knowledge management can improve decision-making quality in medium-sized technology companies. This review asks how the existing literature explains the relationship between artificial intelligence-driven knowledge management and decision-making quality in medium-sized technology companies, and whether knowledge management practices can be understood as the main pathway through which this relationship takes place. The review is guided by four supporting questions. First, what theoretical foundations do classical knowledge management studies provide for understanding organisational knowledge and decision-making? Second, how does recent AI, generative AI and knowledge management literature explain the role of intelligent tools in knowledge acquisition, sharing and application? Third, what contextual insights does regional and emerging-economy implementation evidence add regarding readiness, skills, governance and implementation barriers? Fourth, how can these streams of literature be integrated into a conceptual framework linking AI-driven knowledge management, knowledge management practices, decision-making quality and perceived challenges? 3. Methodology of the Review This paper adopts a purposive critical integrative review. Torraco (2005) and Whittemore and Knafl (2005) provide the main methodological basis for using an integrative review to connect heterogeneous evidence and develop conceptual insight. More general literature-review guidance supports transparent synthesis of fragmented research streams and the development of conceptual contributions (Fisch & Block, 2018; Paul & Criado, 2020; Snyder, 2019; Tranfield et al., 2003). The approach is appropriate here because the research problem sits between knowledge management, artificial intelligence, generative AI and decision-making quality. The review followed a purposive and staged search process. The first stage focused on identifying foundational studies in knowledge management and organisational knowledge. The second stage focused on recent studies linking artificial intelligence, generative AI, knowledge management and decision-making. The third stage focused on regional and emerging-economy implementation evidence that could provide contextual insight into readiness, skills, infrastructure, acceptance and governance. This staged search was used because the paper does not examine one isolated literature area. It brings together theoretical, technological and contextual streams that are usually discussed separately. Arabic-language studies were identified through combinations of English and Arabic search terms in Google Scholar and regional journal portals, followed by backward citation tracing; their bibliographic details were checked against the journal or DOI record. The search used Scopus, Web of Science, Google Scholar and the AIS electronic library. Searches combined terms such as knowledge management, artificial intelligence, generative AI, decision-making quality, organisational knowledge, AI-driven knowledge management and medium-sized companies. Classical studies were also identified through backward citation tracing because many foundational works were published before recent database filters would normally capture them. During final revision, a targeted verification search on large-language-model hallucination identified one additional peer-reviewed survey (Huang et al., 2025). Because this source was added during revision rather than through the original staged screening process, it is reported separately in Table 3. The selection process was guided by conceptual relevance rather than numerical exhaustiveness. Titles and abstracts were reviewed first to remove studies that were unrelated to organisational knowledge, decision-making or AI-supported knowledge work. The remaining studies were then reviewed more closely to assess their relevance to the research problem, their conceptual or empirical contribution and their usefulness for framework development. Studies were retained when they helped explain one of the main components of the review: AI-driven knowledge management, knowledge management practices, decision-making quality, automation and augmentation, generative AI or perceived implementation challenges. Table 3 summarises the search process and retained sources. Table 3 Search transparency summary for the purposive review Database / Source Key search terms Date range Records screened Sources retained Scopus AI, knowledge management, decision-making quality, generative AI, AI governance 1990–2026 ~160 19 Web of Science AI, KM practices, organisational knowledge, human-AI collaboration, decision quality 1990–2026 ~130 14 Google Scholar Classical KM, AIKM, decision-making quality, generative AI, hallucination, AI trustworthiness 1990–2026 ~210 18 AIS electronic library Information systems, knowledge management, AI in organisations, LLMs and organisational systems 1990–2026 ~45 5 Backward citation tracing Foundational KM and organisational knowledge works Foundational works retained regardless of date ~25 5 Targeted revision search LLM hallucination; peer-reviewed survey 2025 Not separately logged 1 Total ~570 62 Note: figures are approximate and reflect the purposive nature of this review. Counts across Scopus, Web of Science and Google Scholar include overlap and should not be read as unique records. Google Scholar was used primarily as a supplementary source for citation-chasing and locating regional or otherwise difficult-to-index material, rather than as a primary bibliographic database. The original staged search retained 61 sources. One additional peer-reviewed source was added during final revision through a targeted hallucination-verification search and is reported separately; the review therefore contains 62 retained sources. The approximate screening total refers to the original purposive search and does not include a separately reconstructed count for the revision search. Table 4 sets out the inclusion and exclusion criteria applied in the review. Table 4 Inclusion and exclusion criteria used in the review Criterion Inclusion criteria Exclusion criteria Topic relevance Studies addressing knowledge management, artificial intelligence, generative AI, decision-making, organisational knowledge or review methodology Studies dealing with AI only as a technical issue, unless directly relevant to an organisational risk or framework component Type of contribution Theoretical, empirical or review-based studies that support the research problem or framework Exclude studies that do not address at least one framework component or provide a usable theoretical, empirical or review contribution. Contextual relevance International, regional and emerging-economy studies relevant to organisations, knowledge work, AI implementation or decision-making Studies focused on unrelated sectors or purely technical applications Methodological clarity Studies with clear argument, method or conceptual contribution Studies lacking sufficient methodological or conceptual clarity Use in synthesis Studies that help explain AIKM, KM practices, decision-making quality, generative AI or perceived challenges Exclude studies whose findings cannot be mapped to AIKM, KM practices, decision quality, generative AI or implementation challenges. Publication period Main search period 1990–2026, with earlier foundational works retained where theoretically necessary Earlier works excluded unless they are foundational to the theoretical framework These criteria were used to maintain the focus of the review and to ensure that the selected studies contributed directly to the synthesis and the development of the proposed framework. The review does not claim to be an exhaustive systematic review. Instead, it uses selected literature from knowledge management, artificial intelligence in organisational contexts, generative AI, decision-making, review methodology and regional implementation evidence to build a conceptual argument. The synthesis procedure followed a thematic logic. Selected studies were not treated only as separate findings. They were grouped according to the role they played in the argument of the paper. Classical knowledge management studies were used to explain the theoretical basis of organisational knowledge. Contemporary AI and generative AI studies were used to explain how intelligent tools may support and reshape knowledge work. Decision-making studies were used to connect knowledge processes with decision quality. Regional and emerging-economy implementation evidence was used to identify contextual barriers and implementation conditions. The selected studies were assessed through four dimensions: methodological clarity, theoretical contribution, relevance to the research problem and originality. The review was then analysed thematically. This allowed the discussion to move from foundations of knowledge management to artificial intelligence and generative AI in knowledge work, and then to contextual evidence on implementation challenges. This procedure allowed the review to move from description toward conceptual integration. 4. Review of the Literature 4.1. Classical Foundations The classical literature provides the theoretical base for this paper. Nonaka and Takeuchi (1995) remain important because their work explains a problem that still appears in many companies. Practical experience is often held by engineers, managers and project staff before it becomes part of organisational knowledge. When this experience is not documented, the organisation may lose part of what it has already learned. Grant (1996) viewed the firm as an institution that brings specialised knowledge together. This view supports the argument that knowledge is not just an administrative record. It is a strategic resource. Choo (1998) also moved the discussion beyond access to information. His work is useful here because it shows that decisions require interpretation and sense-making before information can guide action. Organisational memory and organisational learning studies also support this view. They show that organisations do not retain knowledge through documents alone, but through routines, individuals, culture, structures and stored information. This is important for medium-sized technology companies because practical experience may remain available only when it is connected to organisational routines and not left as personal memory (Huber, 1991; Walsh & Ungson, 1991). The knowledge-based view also helps explain why knowledge should be treated as an organisational capability. Firms create value not only by possessing knowledge, but by combining, transferring and reusing specialised knowledge in ways that support action and adaptation (Kogut & Zander, 1992; Spender, 1996; Teece et al., 1997). Davenport and Prusak (1998) described knowledge as experience, values, context and expert insight. This helps avoid a narrow technical view of artificial intelligence. Alavi and Leidner (2001) later explained knowledge management systems as socio-technical systems that support knowledge creation, storage, transfer and application. Gold et al. (2001) added that technology, structure and culture need to work together. These studies point to the same general lesson. Knowledge has little value when it remains only in individual memory or scattered across files that are difficult to use. Zack (1999) also linked knowledge management with organisational strategy, which supports the view that companies should know which knowledge is most important for their work. Mills and Smith (2011) found that some knowledge management resources—notably organisational structure and knowledge application—were directly related to organisational performance, whereas other resources were not directly related. Although these classical studies provide a strong theoretical foundation, their main limitation is that they were developed before the current use of artificial intelligence in organisational knowledge work. They explain why knowledge creation, sharing, memory and application matter, but they do not explain how intelligent tools may change the way knowledge is searched, retrieved and reused in decision situations. This limitation is important for the present paper because it shows the need to extend classical knowledge management thinking into an AI-supported organisational context. 4.2. Contemporary Studies on Artificial Intelligence and Knowledge Management Contemporary literature updates the classical discussion by showing how artificial intelligence may support knowledge work. Davenport and Ronanki (2018) identified practical organisational uses of artificial intelligence, including automation, cognitive insight and cognitive engagement. Their work is useful, although much of their discussion is based on larger organisations. Other studies have examined the relationship between artificial intelligence and knowledge management more directly. Taherdoost and Madanchian (2023) reviewed how artificial intelligence can support knowledge discovery, storage, retrieval and analysis. Pai et al. (2022) also showed that the value of artificial intelligence in knowledge management depends on the relationship between people and technology. Jarrahi et al. (2023) made a similar point through the idea of human and artificial intelligence partnership in knowledge work. More recent research treats artificial intelligence as part of organisational knowledge work rather than as a purely technical tool. This view is important because AI affects how people search, interpret, organise and use knowledge inside organisations. It also draws attention to the relationship between intelligent systems, human expertise, organisational routines and governance (Berente et al., 2021; Faraj et al., 2018; Jarrahi, 2018; Raisch & Krakowski, 2021; von Krogh, 2018). Other studies add that the relationship between AI and knowledge management should not be separated from tacit knowledge, implementation challenges and the recent development of generative AI. These issues are relevant to the present review because medium-sized technology companies often depend on practical experience that is difficult to capture fully in formal systems (Rezaei, 2025; Sanzogni et al., 2017; Storey, 2025). Nakash and Bolisani (2024) also show that research between knowledge management and artificial intelligence is still developing and needs more focused organisational studies. Recent work has extended these insights into the generative AI era. Dwivedi et al. (2023) provided a multidisciplinary assessment of generative AI’s opportunities and challenges across research, practice and policy, including risks involving accuracy, bias, transparency and credibility. Kirchner et al. (2025) examined generative AI adoption among software developers as knowledge workers, finding that developers valued GenAI for solving simpler programming tasks efficiently and rapidly, while knowledge exchange with fellow programmers was partly—but not entirely—replaced by exchange with GenAI. This pattern is directly relevant to medium-sized technology companies, where informal knowledge exchange is often a primary channel for transmitting project experience and tacit expertise. He and Yang (2026) combined literature analysis with Chinese manufacturing case studies to develop a five-stage GenAI-enhanced knowledge management framework covering acquisition, sharing, integration, application and optimization. Alavi et al. (2024) examined generative AI through a knowledge management lens, identifying opportunities and challenges across knowledge creation, storage, transfer and application, including risks such as AI bias, reduced human socialization and overreliance on AI. The trustworthiness of AI-generated knowledge outputs has emerged as a specific concern. Hallucinations—instances in which generative AI systems produce fluent but nonfactual or unsupported content—create a risk for organisations that use AI-supported knowledge without adequate validation routines (Huang et al., 2025; Xu et al., 2024). Huang et al. (2025) provide a peer-reviewed survey of hallucination principles, causes, detection and mitigation, while Xu et al. (2024) argue that hallucination is an inherent limitation of large language models. Retrieval-augmented generation can improve grounding by connecting a model to external sources, but the recent literature also shows that retrieval quality, source selection and faithful use of retrieved material remain open technical problems with important organisational implications (Fan et al., 2024; Huang & Huang, 2026; Mombaerts et al., 2024). These limitations reinforce the need for provenance tracking, validation procedures and human review. Chau and Xu (2025) identify major organisational issues and challenges in the use of large language models and call for further information-systems research on their business and management impacts. Leoni et al. (2024), based on semi-structured interviews with KM and AI experts from 52 mostly large, private and for-profit organisations, found that AI adoption in knowledge management has both linear and retroactive relationships with organisational decision-making. The study is useful as exploratory evidence about organisational processes, but it does not test decision-making quality as a construct or establish the mediation proposed in this paper; its large-firm sample also limits direct transfer to medium-sized firms. Many contemporary studies explain the potential of artificial intelligence, but they do not always show how medium-sized technology companies can adopt it under limited resources and less formalised knowledge structures. This is an important gap because medium-sized firms may need practical and gradual adoption rather than large-scale transformation programmes. Gelashvili-Luik et al. (2025) conducted a systematic literature review on integrating emerging AI technologies into knowledge management systems, highlighting governance, data quality and organisational readiness among the recurring implementation concerns identified across sectors. These findings are particularly relevant to medium-sized technology companies, which typically lack the specialised teams or formal governance frameworks used by larger corporations to address such concerns. Oldemeyer et al. (2025) similarly reviewed AI implementation in small and medium enterprises and identified lack of knowledge, costs and inadequate infrastructure as the most commonly perceived implementation barriers. Recent studies also show that generative AI changes the skills, governance arrangements and validation routines required for organisational knowledge work. Kaczorowska-Spychalska et al. (2024) describe generative AI as a potential source of change in the knowledge management paradigm, highlighting its capacity to automate tasks and generate insights while also identifying challenges involving data quality, human oversight and ethical considerations. Korzynski et al. (2023) identify prompt engineering as a new digital competence, which is relevant because employees need the ability to question, refine and evaluate AI-supported outputs rather than accept them passively. Wach et al. (2023) further emphasise the risks and controversies surrounding ChatGPT, including misinformation and misuse, which reinforces the need for validation and human review in organisational settings. Studies on retrieval-augmented generation also show that connecting language models to external knowledge sources can improve grounding but does not remove technical problems of source selection and evaluation; in organisational use, these limitations create a continued need for governance (Fan et al., 2024; Huang & Huang, 2026; Mombaerts et al., 2024). These studies strengthen the present paper’s argument that the value of AI-driven knowledge management depends not only on retrieval capability, but also on the organisational practices that make AI-supported knowledge traceable, verifiable and decision-ready. The contemporary literature is useful because it shows that artificial intelligence can support knowledge discovery, retrieval, analysis and decision support. However, much of this literature still treats AI adoption as a general organisational or technological issue. It does not always explain the knowledge management routines that must exist before AI can produce value. It also gives limited attention to medium-sized technology companies, where knowledge may be rich in practical terms but less formally organised. This leaves an important gap between the promise of AI and the organisational conditions needed for its effective use. 4.3. Regional and Emerging-Economy Implementation Evidence Regional and emerging-economy implementation evidence adds a useful contextual layer to the review. Abu Al-Nasr (2021) discussed knowledge management and knowledge-based management in Arab institutions, which is useful for linking the topic with regional organisational contexts. More recent regional studies address AI in organisational and educational decision and knowledge-management settings (Al Azzam & Al Dafra, 2023; Al-Dosari & Al-Nouh, 2024; Al-Qarni, 2024), while Masameh et al. (2025) provide complementary public-sector evidence on knowledge management and institutional development. The value of this literature for the present paper is mainly contextual. It draws attention to practical implementation conditions such as digital readiness, skills, infrastructure, acceptance, governance and institutional support. These issues are important because AI-driven knowledge management cannot work effectively through technology alone. It also depends on the organisational environment in which knowledge is documented, reviewed, shared and used. At the same time, much of this regional evidence remains concentrated in public and educational sectors rather than medium-sized technology companies. It also tends to discuss artificial intelligence and decision-making as a relatively direct relationship, without giving enough attention to knowledge management practices as the pathway through which AI may influence decision-making quality. The present paper therefore uses this literature as implementation evidence rather than as the main theoretical foundation. It supports the argument that AI-supported knowledge work is shaped by readiness, governance and human capability, especially in organisations where formal knowledge systems are still developing. The relevance of this evidence to medium-sized technology companies can be justified through the similarity of the organisational bottlenecks involved. Public and educational institutions often face bureaucratic routines, fragmented data, weak governance arrangements and uneven digital readiness when they introduce AI-supported systems. Medium-sized technology firms may operate in a more commercial and project-based environment, but they can face comparable structural constraints when knowledge is dispersed across departments, individual expertise, supplier files, project records and client histories. For this reason, regional implementation evidence is used here not as direct sector evidence, but as a theoretical bridge for understanding how data silos, governance gaps and human capability constraints may also shape AI-driven knowledge management in medium-sized technology companies. 4.4. Summary of the Literature Streams Table 5 Summary of the literature streams used in the review Literature stream What it explains well Main limitation Use in this paper Classical knowledge management Knowledge creation, sharing, strategy and socio-technical systems Most works predate current artificial intelligence tools Provides the theoretical foundation Contemporary international studies Artificial intelligence in knowledge work and decision support Often focused on large organisations or broad digital transformation Updates the discussion with recent intelligent tools Regional and emerging-economy implementation evidence Readiness, skills, infrastructure, governance and implementation conditions Often focused on public and educational sectors rather than medium-sized technology companies Adds contextual evidence on readiness, governance and implementation challenges Table 5 shows that the three streams are complementary. Classical studies explain why knowledge matters. Contemporary studies show how artificial intelligence and generative AI may support and reshape knowledge processes. Regional and emerging-economy implementation evidence shows that digital readiness, skills, governance and institutional challenges matter in organisational settings. The gap lies in bringing these streams together for medium-sized technology companies. 5. Synthesis and Research Gap The reviewed literature can be synthesised around one main point. AI can support knowledge work, but it does not remove the need for knowledge management. Earlier knowledge systems mainly helped organisations store, classify and retrieve what had already been documented. Generative AI adds a new issue because it can also summarise, combine and produce knowledge-like outputs. This makes AI-supported knowledge work more powerful, but also more dependent on validation, provenance, governance and human review. Classical knowledge management studies show that knowledge must be created, shared, stored and applied before it can become useful for organisational action. Contemporary AI studies show that intelligent tools can support retrieval, classification, summarisation, analysis and decision support. Decision-making research highlights the role of information and decision processes (Citroen, 2011; Dean & Sharfman, 1996), examines decision speed in high-velocity environments (Eisenhardt, 1989), documents the rapid adoption of data-driven decision-making (Brynjolfsson & McElheran, 2016), and develops alternative human–AI decision structures intended to benefit organisational decision quality (Shrestha et al., 2019). Regional and emerging-economy implementation evidence also points to skills, infrastructure, readiness, governance and acceptance as practical implementation conditions. Taken together, these streams suggest that AI-driven knowledge management should not be examined only as a technical capability. It should be examined as a socio-technical knowledge process. If a company has weak documentation, scattered files and poor sharing routines, AI may only make weak material easier to find or reproduce. If the company has stronger knowledge practices, AI can make previous experience more visible, searchable and usable. In the case of generative AI, this point becomes more important because generated outputs may look coherent even when they require checking against reliable organisational sources. The research gap can therefore be stated as follows: existing literature has not yet explained how generative AI changes the knowledge management practices that matter most for decision-making quality in medium-sized technology companies. Prior research focused on whether AI improves access to knowledge. The present paper argues that the more important and less examined question is whether organisations can make AI-supported knowledge trustworthy enough to use in decisions. This requires a shift in theoretical focus from retrieval-oriented knowledge management practices to trustworthiness-oriented ones, specifically provenance, validation, governance and human review. This gap is especially consequential for medium-sized technology companies, where formal knowledge governance is often limited and where the risks of acting on ungoverned AI outputs are high. These firms may hold valuable project experience, supplier history, technical expertise and client knowledge, but this knowledge may remain dispersed across files, people and previous decisions. For such companies, AI-driven knowledge management may create value only when it is connected to clear practices for documenting, validating, retrieving, sharing and reviewing knowledge before decisions are made. This synthesis is informed by several converging lines of research. Shrestha et al. (2019) develop a framework for combining human and AI-based decision-making across different contingency conditions, with the aim of benefiting organisational decision quality. Berente et al. (2021) frame artificial intelligence as a distinct organisational management challenge rather than a technology that can be adopted passively. Raisch and Krakowski (2021) conceptualise automation and augmentation as interdependent rather than mutually exclusive, highlighting the need to manage their paradoxical relationship. Together, these studies motivate—but do not empirically establish—the mediating logic of the present framework: AI-driven knowledge management is expected to create value for decisions through knowledge management practices that structure how knowledge is captured, validated and used, rather than through technical deployment alone. 6. Proposed Conceptual Framework The proposed framework is built on the synthesis above. It links four main components: AI-driven knowledge management, knowledge management practices, decision-making quality and perceived challenges. The framework does not treat AI as an automatic solution. It treats AI as an enabling condition whose value depends on how it is connected to organisational knowledge practices. The framework is informed by the automation-augmentation view of artificial intelligence. AI can automate some knowledge tasks, such as searching, classifying, summarising and comparing previous cases. At the same time, AI can augment human work by helping managers and employees interpret knowledge, notice patterns and review previous experience before decisions are made. This distinction is important for the present paper because decision-making quality in medium-sized technology companies cannot be improved by automation alone. It also requires human judgement, validation and contextual understanding. Generative AI makes this issue more important. It can produce knowledge-like outputs that appear useful and coherent, but these outputs still need to be checked against reliable organisational sources. For this reason, the framework gives knowledge management practices a central position. Practices such as documentation, validation, provenance, retrieval, sharing, application, governance and human review are treated as the pathway through which AI-supported knowledge can become useful for decision-making. The position of the variables in the framework follows this logic. AI-driven knowledge management is placed as the independent variable because it represents the organisational use of intelligent tools to acquire, organise, retrieve, generate, summarise and apply knowledge. Knowledge management practices are placed as the mediating variable because the value of AI is expected to pass through organisational routines rather than through technical presence alone. Decision-making quality is placed as the dependent variable because it represents the expected organisational outcome of better knowledge use. Perceived challenges are included as moderating conditions because weak documentation, limited skills, poor data quality, resistance, unclear governance and overreliance on AI may reduce the expected value of AI-supported knowledge work. Figure 1 presents the proposed conceptual framework. The main path moves from AI-driven knowledge management to knowledge management practices and then to decision-making quality. A secondary direct path from AI-driven knowledge management to decision-making quality is also shown. This path recognises that AI tools may sometimes support decisions directly through search, summarisation or analytical support. However, the framework assumes that this direct path is weaker and less stable than the mediated path through knowledge management practices. Perceived challenges are shown as moderating conditions because they may weaken both the use of AI in knowledge practices and the translation of those practices into decision-making quality. The logic of the framework is therefore based on partial rather than full mediation. AI-driven knowledge management may have some direct value for decision-making, but its more reliable contribution is expected to occur through knowledge management practices. This is especially important in medium-sized technology companies, where valuable knowledge may exist but remain dispersed across people, documents, projects and previous decisions. Figure 1 Proposed conceptual framework Note. Solid arrows indicate direct or mediated paths. Dashed moderation paths indicate that perceived challenges may weaken both the AI-driven knowledge management → knowledge management practices path and the knowledge management practices → decision-making quality path. 6.1. Research Propositions Based on the proposed framework, the review develops six conceptual propositions for future empirical testing. These propositions are not presented as tested hypotheses in this paper. They are derived from the automation-augmentation view of AI and from the argument that provenance, validation, governance and human review are the new core knowledge management practices in AI-supported organisations. They are intended to clarify how AI-driven knowledge management may influence decision-making quality through these practices, and how perceived challenges may weaken these relationships. AI-driven knowledge management can support knowledge practices by automating and augmenting parts of knowledge work. It may help organisations search previous records, classify documents, summarise project experience and retrieve relevant knowledge more quickly. In medium-sized technology companies, this can be useful because valuable knowledge is often dispersed across emails, reports, supplier files, technical documents and individual experience. However, AI contributes to knowledge management only when it is connected to organisational routines that make knowledge visible, usable and reviewable. Proposition 1: AI-driven knowledge management positively influences knowledge management practices in medium-sized technology companies. Knowledge management practices are expected to influence decision-making quality because decisions depend on the quality and usability of the knowledge available at the time of decision. When knowledge is documented, validated, shared and applied through clear routines, managers and employees can make decisions that are better informed, more timely and more consistent with organisational objectives. In this sense, knowledge management practices form the organisational route through which knowledge becomes useful for decision-making. Proposition 2: Knowledge management practices positively influence decision-making quality. The relationship between AI-driven knowledge management and decision-making quality is expected to operate mainly through knowledge management practices. AI may improve access to knowledge, but access alone is not enough. Knowledge must be checked, interpreted, traced to its source and applied to the decision context. This is especially important with generative AI, because generated outputs may appear coherent even when they require validation against reliable organisational sources. Proposition 3: Knowledge management practices mediate the relationship between AI-driven knowledge management and decision-making quality. A secondary direct effect may also exist. AI tools can sometimes support decision-making directly by providing faster search, summarisation, comparison of previous cases or analytical support. However, this direct effect is expected to be weaker and less stable than the mediated effect through knowledge management practices, because direct AI outputs still require human judgement and organisational validation before they can safely support decisions. Proposition 4: AI-driven knowledge management may have a secondary direct positive influence on decision-making quality, but this effect is expected to be weaker than the mediated pathway through knowledge management practices. Perceived challenges may weaken the relationship between AI-driven knowledge management and knowledge management practices. Poor data quality, weak documentation, limited technical skills, unclear governance and resistance to change can reduce the ability of AI tools to support knowledge work. In such conditions, AI may retrieve or generate outputs faster, but those outputs may not become reliable organisational knowledge. Proposition 5: Perceived challenges weaken the relationship between AI-driven knowledge management and knowledge management practices. Perceived challenges may also weaken the relationship between knowledge management practices and decision-making quality. Even when knowledge practices exist, their value may be reduced if employees do not trust the system, if governance rules are unclear, if knowledge is not updated, or if decision-makers over-rely on AI outputs without sufficient human review. These challenges may reduce the ability of knowledge practices to support timely, evidence-based and contextually appropriate decisions. Proposition 6: Perceived challenges weaken the relationship between knowledge management practices and decision-making quality. 7. Discussion The review supports a balanced understanding of AI-driven knowledge management. AI can make knowledge easier to search, summarise and compare, but this does not automatically improve decision-making quality. Its value appears when connected to knowledge management practices that make organisational knowledge visible, reliable and usable — which is why the proposed framework places knowledge management practices as the main pathway between AI-driven knowledge management and decision-making quality. In the generative AI context, outputs that appear clear and coherent may still need to be checked against reliable organisational sources. Validation, provenance, governance and human review are therefore the core knowledge management practices through which AI-supported knowledge becomes trustworthy enough to use in decisions. The framework also reflects the automation-augmentation view of AI. Some AI functions may automate knowledge work, such as classification, retrieval and summarisation. Other functions may augment human work by helping managers and employees interpret previous cases, compare alternatives and prepare for decisions. The present review argues that decision-making quality depends more on augmentation than automation alone. Medium-sized technology companies may gain value from automation, but reliable decisions still require human judgement and contextual understanding. This argument extends classical knowledge management theory. Classical knowledge management studies were written before the current rise of AI, but their ideas remain useful because knowledge still needs to be captured, shared, interpreted and applied (Alavi & Leidner, 2001; Davenport & Prusak, 1998; Nonaka & Takeuchi, 1995). What has changed is that AI, especially generative AI, can now participate in the production and reformulation of knowledge-like outputs. This makes the old knowledge management problem more complex rather than less important. The framework also differs from studies that treat AI mainly as a direct driver of decision quality or performance. A direct path may exist because AI tools can support search, summarisation and analysis. However, this direct effect is expected to be weaker and less stable than the mediated path through knowledge management practices. For medium-sized technology companies, the main issue is not only whether AI tools are available. It is whether the organisation has the routines needed to document, validate, retrieve, share and review knowledge before decisions are made. Regional and emerging-economy implementation evidence adds an important reminder. AI-supported knowledge work is shaped by readiness, skills, infrastructure, governance and institutional context. These issues matter for medium-sized technology companies because they may operate with limited formal knowledge systems and fewer specialised data or AI teams than large corporations. The framework therefore treats perceived challenges as active conditions that can weaken the value of AI-driven knowledge management. 7.1. Implementation Challenges Implementation challenges are central to the proposed framework. Data quality is the first concern: if project records, supplier information and previous decisions are incomplete or poorly organised, AI tools may retrieve information faster without improving its usefulness. In the case of generative AI, poor data quality may also lead to outputs that appear coherent but are not reliable enough for decision-making. Skills are also important. Employees need enough digital awareness to use AI-supported systems correctly, interpret results and recognise their limits. Without these skills, AI may be treated either as a complete substitute for professional judgement or as a tool that employees avoid, both of which reduce its value. Governance is another challenge. Medium-sized technology companies need clear rules about who can add, update, approve and use knowledge inside the system. They also need rules for checking AI-generated summaries and tracing important outputs back to reliable organisational sources. Without such governance, the knowledge base may become inconsistent and difficult to trust. Resistance to change may also weaken adoption when employees feel their practical experience is being replaced rather than supported. AI implementation should therefore be presented as a way to preserve practical knowledge, not remove human expertise. Human judgement remains essential throughout: AI can support retrieval and analysis, but it cannot carry organisational responsibility for the final decision. Managers and specialists must still assess whether previous knowledge fits the new case, client requirement or regulatory context. A further challenge is over-reliance on AI outputs. When AI-generated answers appear clear and confident, decision-makers may give them more trust than they deserve. This risk is higher when the source of the output is not clear or when employees do not check the information against reliable organisational records. Provenance, validation and human review are therefore necessary safeguards for AI-driven knowledge management. Technical research characterises hallucination as a persistent reliability problem in large language models: Xu et al. (2024) argue that hallucination is an inherent limitation, while Huang et al. (2025) synthesise the main causes, detection approaches and mitigation strategies. These sources do not by themselves establish organisational failure rates; rather, they show why fluent AI outputs require verification before they are treated as organisational knowledge. For medium-sized technology companies without formal AI governance teams, this creates a specific risk: confident AI outputs may be accepted as reliable organisational knowledge without adequate human verification. Dwivedi et al. (2023) likewise emphasise that generative AI introduces substantive accuracy, transparency, ethical and organisational risks, reinforcing the need for careful human validation when such outputs are used in knowledge work. These concerns reinforce the need for provenance tracking and structured human review as practical governance mechanisms, not merely theoretical safeguards. The need for human review is also supported by recent evidence on human-AI collaboration. Vaccaro et al. (2024) show that human-AI combinations do not automatically outperform humans or AI alone, and that performance depends on task type and the way collaboration is structured. This finding is important for AI-driven knowledge management because it suggests that human oversight must be designed into the knowledge process rather than assumed. Reuel and Undheim (2024) similarly argue that generative AI requires adaptive governance because its capabilities, uses and risks change quickly. For medium-sized technology companies, this means that governance cannot be treated only as a formal policy. It must operate through practical routines for checking outputs, assigning responsibility, tracing sources and deciding when human judgement should override AI-supported recommendations. Hosanagar and Ahn (2024), in a creative-writing experiment, likewise show that collaboration design matters: configurations that preserved substantive human input produced higher quality and satisfaction than designs that limited humans largely to confirming AI output. Although the task context differs from organisational knowledge work, the result supports treating role design as an empirical question rather than assuming that more AI involvement is always better. 8. Theoretical Implications This review offers four theoretical implications. First, it extends classical knowledge management theory by showing that AI-driven tools do not reduce the need for knowledge management practices. In the context of generative AI, these practices become more important because knowledge may now be summarised, combined and reformulated by intelligent systems. Validation, provenance, governance and human review therefore become central parts of AI-supported knowledge management. Second, the paper contributes to AI and knowledge management literature by positioning knowledge management practices as the main pathway through which AI-driven knowledge management may influence decision-making quality. The framework follows a partial mediation logic. AI may have some direct value through search, summarisation and analytical support, but its more reliable contribution is expected to occur through knowledge management practices. Third, the paper adds to decision-making research by linking decision-making quality to the condition of organisational knowledge. Better decisions depend not only on access to AI tools, but also on whether the knowledge used is reliable, traceable, timely and reviewed in context. This is especially important for medium-sized technology companies, where useful knowledge may be practical, project-based and dispersed across people, files and previous decisions. Fourth, the paper contributes to the growing body of work on AI governance in knowledge-intensive organisations. Rezaei (2025) identified a broad set of technological, organisational and ethical challenges associated with AI-enabled knowledge management. The study highlights job security and privacy as prominent cross-process concerns, while transparency, accountability and explainability are also identified as important implementation issues influencing trust in AI-supported knowledge systems. These concerns are amplified in the generative AI context because outputs may be produced without clear attribution to verifiable organisational sources. The present framework addresses this by positioning provenance and governance as core knowledge management practices, rather than as compliance add-ons. Gelashvili-Luik et al. (2025) reached a complementary conclusion, finding that successful AI-enabled knowledge management depends on strong leadership commitment, adaptable governance structures, context-sensitive technology selection and an appropriate balance between automation and human oversight. Together, these contributions suggest that the theoretical framework proposed in the present paper is consistent with emerging research on what makes AI-supported knowledge work reliable and organisationally useful. 9. Practical Implications For managers, the framework suggests beginning with a specific knowledge risk rather than with the purchase of an AI tool. In a 120-person technology firm without a dedicated data team, a workable first step is to choose one bounded repository—such as completed project files, supplier assessments or technical incident reports—assign an owner, define which documents are authoritative, and record the source and date of every AI-generated summary. The purpose is not merely to make search faster, but to ensure that an employee can trace a recommendation back to the organisational evidence on which it relies. A simple provenance routine can operate through four controls. First, the AI output should display or attach the source documents used. Second, a named subject-matter reviewer should verify material claims against those sources. Third, the reviewer should record approval, correction or rejection in the project record. Fourth, any approved summary should carry a review date and an expiry or revalidation point. For high-impact decisions—such as regulatory compliance, safety, client commitments or major expenditure—the accountable manager, not the AI user alone, should sign off before the output enters the decision process. Governance should also specify who may upload, update, approve and reuse organisational knowledge; which categories of information may not be entered into external AI systems; and when professional judgement must override an AI-supported recommendation. These controls respond to the finding that human–AI combinations do not automatically outperform the better of humans or AI working alone (Vaccaro et al., 2024). The framework therefore favours selective augmentation, not an assumption that collaboration is always superior: human review adds value only when roles, expertise, escalation rules and accountability are deliberately designed. 10. Limitations and Future Research This review has several limitations. It is a purposive critical integrative review and does not claim to be an exhaustive systematic review. Screening, selection and synthesis were carried out by the author. Fully independent dual screening was not undertaken, which may introduce selection and interpretive bias; explicit inclusion and exclusion criteria were used to reduce it. The inclusion of public-sector and educational evidence provides contextual insight, but transferability to medium-sized technology companies remains partly analogical and should be tested empirically. The proposed framework is conceptual and has not yet been tested. Future research should test the framework with data from medium-sized technology companies. Quantitative studies could use structural equation modelling to examine the mediating role of knowledge management practices and the moderating role of perceived challenges; established procedural-rationality measures, including those developed by Dean and Sharfman (1996), provide a starting point for operationalising decision-making quality. The perceived-challenges construct is intentionally broad at this conceptual stage, but it combines data quality, documentation, skills, governance, resistance and overreliance, which have different causal logics. Empirical studies should therefore disaggregate these dimensions before testing moderation on the two framework paths. Qualitative and mixed-methods studies could then explain how managers and technical employees apply provenance, validation and human-review routines in daily work. Future work may also examine additional conditions such as leadership support, organisational culture, data governance and the risks of over-reliance on generative AI. These factors may influence whether AI-driven knowledge management becomes a real organisational capability or remains only a technical tool. Table 6 summarises the suggested directions for future research. Table 6 Suggested directions for future research Future research area Suggested method Testing how provenance, validation, governance and human review shape the AIKM → KMP → decision-making pathway Structural equation modelling (PLS-SEM) with separate measures for trustworthiness practices Examining the moderating role of perceived challenges Quantitative survey with moderation analysis Understanding employee acceptance of AI-supported knowledge systems Qualitative interviews Comparing AIKM adoption in medium-sized and large firms Comparative case study Measuring decision-making quality in technology sectors Survey-based empirical study Studying governance, provenance and ethical risks of generative AI in AIKM Mixed-methods research Operationalising decision-making quality and framework propositions Survey measures based on procedural rationality, evidence use, timeliness and goal consistency 11. Conclusion This review concludes that AI-driven knowledge management can support decision-making quality, but not as an isolated technical solution. Its value depends on whether AI tools are connected to knowledge management practices that make organisational knowledge visible, reliable, traceable and usable. This is especially important in medium-sized technology companies, where useful knowledge is often practical, project-based and dispersed across people, files and previous decisions. The review also shows that generative AI changes the conditions of knowledge work. AI can now summarise, combine and produce knowledge-like outputs, but these outputs still require validation, provenance, governance and human review. For this reason, knowledge management practices remain the main pathway through which AI-driven knowledge management may influence decision-making quality. This logic is reflected throughout the framework in the central role assigned to provenance, validation, governance and human review as organisational safeguards rather than optional additions. The proposed framework therefore treats knowledge management practices as a mediating mechanism and perceived challenges as limiting conditions. It also recognises a secondary direct path from AI-driven knowledge management to decision-making quality, but this path is expected to be weaker than the mediated pathway through organised knowledge practices. The paper provides a conceptual basis for future empirical testing in medium-sized technology companies and contributes to a more cautious understanding of AI-supported decision-making. Acknowledgements The author thanks Prof. Dr. Ibrahim Al Souleiman for supervisory guidance and for comments on earlier drafts of this manuscript. Responsibility for the argument, the selection and interpretation of sources, and any remaining errors rests with the author alone. Generative AI Use Statement Generative AI tools were used to support the readability, language refinement and text editing of the manuscript. The author developed the research question, conceptual framework and scholarly argument, appraised and interpreted the literature, and made all final academic decisions. The author reviewed and verified the final content and takes full responsibility for the accuracy and integrity of the manuscript. Funding Statement This research received no external funding. Conflict of Interest Statement The author discloses that Dr. Ibrahim Al Souleiman, acknowledged for supervisory guidance on earlier drafts, has an active role in directing operations and global strategy for the U7Y Academic Journal. No other conflicts of interest are declared. Data Availability Statement This article is based on a critical integrative review of published literature. No new empirical dataset was generated or analysed for this paper. Ethics Approval Statement This study did not involve human participants, personal data collection or experimental procedures. Therefore, formal ethics approval was not required. Originality Statement This manuscript is original, not previously published, and not currently under review elsewhere. References Abu Al-Nasr, M. M. (2021). Idarat al-ma‘rifah wa-al-idarah bi-al-ma‘rifah [Knowledge management and management by knowledge]. The Arab Journal for Informatic and Information Security, 2(4), 75–94. https://doi.org/10.21608/jinfo.2021.183902 Alavi, M., & Leidner, D. E. (2001). Review: Knowledge management and knowledge management systems: Conceptual foundations and research issues. MIS Quarterly, 25(1), 107–136. https://doi.org/10.2307/3250961 Alavi, M., Leidner, D. E., & Mousavi, R. (2024). A knowledge management perspective of generative artificial intelligence. Journal of the Association for Information Systems, 25(1), 1–12. https://doi.org/10.17705/1jais.00859 Al Azzam, S., & Al Dafra, F. (2023). Athar tatbiq al-dhaka’ al-istina‘i ‘ala jawdat ittikhadh al-qararat fi Imarat Mintaqat ‘Asir khilala waba’ COVID-19 [The impact of the application of artificial intelligence on the quality of decision-making in the Emirate of Asir Region during the COVID-19 epidemic]. Arab Journal of Administration, 43(4), 347–360. https://doi.org/10.21608/aja.2022.116479.1195 Al-Dosari, M. F. B., & Al-Nouh, A. S. (2024). Waqi‘ tawzif al-dhaka’ al-istina‘i li-tatbiq ‘amaliyyat idarat al-ma‘rifah bi-Wizarat al-Ta‘lim fi al-Mamlakah al-‘Arabiyyah al-Su‘udiyyah [The reality of employing artificial intelligence to implement knowledge management processes at the Ministry of Education in Saudi Arabia]. Journal of Arts, Literature, Humanities and Social Sciences, (113), 316–339. https://doi.org/10.33193/JALHSS.113.2024.1257 Al-Qarni, M. A. (2024). Qiyas athar al-dhaka’ al-istina‘i fi tahsin ‘amaliyyat ittikhadh al-qarar bi-madaris al-ta‘lim al-‘amm fi al-Mamlakah al-‘Arabiyyah al-Su‘udiyyah [Measuring the impact of artificial intelligence on improving decision-making processes in public education schools in Saudi Arabia]. Journal of the Faculty of Education – Tanta University, 90(4), 86–151. https://doi.org/10.21608/mkmgt.2024.328024.1826 Berente, N., Gu, B., Recker, J., & Santhanam, R. (2021). Special issue editor’s comments: Managing artificial intelligence. MIS Quarterly, 45(3), 1433–1450. https://doi.org/10.25300/MISQ/2021/16274 Brynjolfsson, E., & McElheran, K. (2016). The rapid adoption of data-driven decision-making. American Economic Review, 106(5), 133–139. https://doi.org/10.1257/aer.p20161016 Chau, M., & Xu, J. (2025). An IS research agenda on large language models: Development, applications, and impacts on business and management. ACM Transactions on Management Information Systems, 16(1), Article 1, 1–11. https://doi.org/10.1145/3713032 Choo, C. W. (1998). The knowing organization: How organizations use information to construct meaning, create knowledge, and make decisions. Oxford University Press. Citroen, C. L. (2011). The role of information in strategic decision-making. International Journal of Information Management, 31(6), 493–501. https://doi.org/10.1016/j.ijinfomgt.2011.02.005 Davenport, T. H., & Prusak, L. (1998). Working knowledge: How organizations manage what they know. Harvard Business School Press. Davenport, T. H., & Ronanki, R. (2018). Artificial intelligence for the real world. Harvard Business Review, 96(1), 108–116. https://hbr.org/2018/01/artificial-intelligence-for-the-real-world Dean, J. W., Jr., & Sharfman, M. P. (1996). Does decision process matter? A study of strategic decision-making effectiveness. Academy of Management Journal, 39(2), 368–396. https://doi.org/10.5465/256784 Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., Albanna, H., Albashrawi, M. A., Al-Busaidi, A. S., Balakrishnan, J., Barlette, Y., Basu, S., Bose, I., Brooks, L., Buhalis, D., … Wright, R. (2023). Opinion paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 71, Article 102642. https://doi.org/10.1016/j.ijinfomgt.2023.102642 Eisenhardt, K. M. (1989). Making fast strategic decisions in high-velocity environments. Academy of Management Journal, 32(3), 543–576. https://doi.org/10.5465/256434 European Commission. (2003). Commission Recommendation 2003/361/EC of 6 May 2003 concerning the definition of micro, small and medium-sized enterprises. Official Journal of the European Union, L 124, 36–41. https://eur-lex.europa.eu/eli/reco/2003/361/oj Fan, W., Ding, Y., Ning, L., Wang, S., Li, H., Yin, D., Chua, T.-S., & Li, Q. (2024). A survey on RAG meeting LLMs: Towards retrieval-augmented large language models. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (pp. 6491–6501). Association for Computing Machinery. https://doi.org/10.1145/3637528.3671470 Faraj, S., Pachidi, S., & Sayegh, K. (2018). Working and organizing in the age of the learning algorithm. Information and Organization, 28(1), 62–70. https://doi.org/10.1016/j.infoandorg.2018.02.005 Fisch, C., & Block, J. (2018). Six tips for your (systematic) literature review in business and management research. Management Review Quarterly, 68(2), 103–106. https://doi.org/10.1007/s11301-018-0142-x Gelashvili-Luik, T., Vihma, P., & Pappel, I. (2025). Navigating the AI revolution: Challenges and opportunities for integrating emerging technologies into knowledge management systems. Systematic literature review. Frontiers in Artificial Intelligence, 8, 1595930. https://doi.org/10.3389/frai.2025.1595930 Gold, A. H., Malhotra, A., & Segars, A. H. (2001). Knowledge management: An organizational capabilities perspective. Journal of Management Information Systems, 18(1), 185–214. https://doi.org/10.1080/07421222.2001.11045669 Grant, R. M. (1996). Toward a knowledge‐based theory of the firm. Strategic Management Journal, 17(S2), 109–122. https://doi.org/10.1002/smj.4250171110 He, Q., & Yang, Z. (2026). Generative AI-driven knowledge management in manufacturing firms: A five-stage framework for dynamic knowledge optimization and digital innovation. Journal of Knowledge Management, 30(4), 1447–1467. https://doi.org/10.1108/JKM-03-2025-0418 Hosanagar, K., & Ahn, D. (2024). Designing human and generative AI collaboration. arXiv preprint. https://doi.org/10.48550/arXiv.2412.14199 Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., & Liu, T. (2025). A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions on Information Systems, 43(2), Article 42, 1–55. https://doi.org/10.1145/3703155 Huang, Y., & Huang, J. (2026). A survey on retrieval-augmented text generation for large language models. ACM Computing Surveys, 58(12), Article 300, 1–38. https://doi.org/10.1145/3805774 Huber, G. P. (1991). Organizational learning: The contributing processes and the literatures. Organization Science, 2(1), 88–115. https://doi.org/10.1287/orsc.2.1.88 Jarrahi, M. H. (2018). Artificial intelligence and the future of work: Human-AI symbiosis in organizational decision making. Business Horizons, 61(4), 577–586. https://doi.org/10.1016/j.bushor.2018.03.007 Jarrahi, M. H., Askay, D., Eshraghi, A., & Smith, P. (2023). Artificial intelligence and knowledge management: A partnership between human and AI. Business Horizons, 66(1), 87–99. https://doi.org/10.1016/j.bushor.2022.03.002 Kaczorowska-Spychalska, D., Kotula, N., Mazurek, G., & Sułkowski, Ł. (2024). Generative AI as source of change of knowledge management paradigm. Human Technology, 20(1), 131–154. https://doi.org/10.14254/1795-6889.2024.20-1.7 Kirchner, K., Bolisani, E., Kassaneh, T. C., Scarso, E., & Taraghi, N. (2025). Generative AI meets knowledge management: Insights from software development practices. Knowledge and Process Management, 32(4), 223–235. https://doi.org/10.1002/kpm.70004 Kogut, B., & Zander, U. (1992). Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science, 3(3), 383–397. https://doi.org/10.1287/orsc.3.3.383 Korzynski, P., Mazurek, G., Krzypkowska, P., & Kurasinski, A. (2023). Artificial intelligence prompt engineering as a new digital competence: Analysis of generative AI technologies such as ChatGPT. Entrepreneurial Business and Economics Review, 11(3), 25–37. https://doi.org/10.15678/EBER.2023.110302 Leoni, L., Ardolino, M., El Baz, J., Gueli, G., & Bacchetti, A. (2022). The mediating role of knowledge management processes in the effective use of artificial intelligence in manufacturing firms. International Journal of Operations & Production Management, 42(13), 411–437. https://doi.org/10.1108/IJOPM-05-2022-0282 Leoni, L., Gueli, G., Ardolino, M., Panizzon, M., & Gupta, S. (2024). AI-empowered KM processes for decision-making: Empirical evidence from worldwide organisations. Journal of Knowledge Management, 28(11), 320–347. https://doi.org/10.1108/JKM-03-2024-0262 Masameh, M., Al-Qalash, O. A. G., & Shaheen, S. K. M. (2025). Dawr idarat al-ma‘rifah fi bina’ al-mu’assasat al-hukumiyyah wa-tatwiraha: Muraja‘ah ‘ilmiyyah [The role of knowledge management in building and developing government institutions: A scientific review]. Arab International Journal of Knowledge Management, 4(3), 103–186. https://doi.org/10.21608/aikm.2025.347779.1101 Mills, A. M., & Smith, T. A. (2011). Knowledge management and organizational performance: A decomposed view. Journal of Knowledge Management, 15(1), 156–171. https://doi.org/10.1108/13673271111108756 Mombaerts, L., Ding, T., Banerjee, A., Felice, F., Taws, J., & Borogovac, T. (2024). Meta knowledge for retrieval augmented large language models. arXiv preprint. https://doi.org/10.48550/arXiv.2408.09017 Nakash, M., & Bolisani, E. (2024). Knowledge management meets artificial intelligence: A systematic review and future research agenda. European Conference on Knowledge Management, 25(1), 544–552. https://doi.org/10.34190/eckm.25.1.2443 Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese companies create the dynamics of innovation. Oxford University Press. https://doi.org/10.1093/oso/9780195092691.001.0001 Oldemeyer, L., Jede, A., & Teuteberg, F. (2025). Investigation of artificial intelligence in SMEs: A systematic review of the state of the art and the main implementation challenges. Management Review Quarterly, 75(2), 1185–1227. https://doi.org/10.1007/s11301-024-00405-4 Pai, R. Y., Shetty, A., Shetty, A. D., Bhandary, R., Shetty, J., Nayak, S., Dinesh, T. K., & D’souza, K. J. (2022). Integrating artificial intelligence for knowledge management systems – synergy among people and technology: A systematic review of the evidence. Economic Research-Ekonomska Istraživanja, 35(1), 7043–7065. https://doi.org/10.1080/1331677X.2022.2058976 Paul, J., & Criado, A. R. (2020). The art of writing literature review: What do we know and what do we need to know? International Business Review, 29(4), Article 101717. https://doi.org/10.1016/j.ibusrev.2020.101717 Raisch, S., & Krakowski, S. (2021). Artificial intelligence and management: The automation–augmentation paradox. Academy of Management Review, 46(1), 192–210. https://doi.org/10.5465/amr.2018.0072 Reuel, A., & Undheim, T. A. (2024). Generative AI needs adaptive governance. arXiv preprint. https://doi.org/10.48550/arXiv.2406.04554 Rezaei, M. (2025). Artificial intelligence in knowledge management: Identifying and addressing the key implementation challenges. Technological Forecasting and Social Change, 217, 124183. https://doi.org/10.1016/j.techfore.2025.124183 Sanzogni, L., Guzman, G., & Busch, P. (2017). Artificial intelligence and knowledge management: Questioning the tacit dimension. Prometheus, 35(1), 37–56. https://doi.org/10.1080/08109028.2017.1364547 Shrestha, Y. R., Ben-Menahem, S. M., & von Krogh, G. (2019). Organizational decision-making structures in the age of artificial intelligence. California Management Review, 61(4), 66–83. https://doi.org/10.1177/0008125619862257 Snyder, H. (2019). Literature review as a research methodology: An overview and guidelines. Journal of Business Research, 104, 333–339. https://doi.org/10.1016/j.jbusres.2019.07.039 Spender, J.-C. (1996). Making knowledge the basis of a dynamic theory of the firm. Strategic Management Journal, 17(S2), 45–62. https://doi.org/10.1002/smj.4250171106 Storey, V. C. (2025). Knowledge management in a world of generative AI: Impact and implications. ACM Transactions on Management Information Systems, 16(3), Article 26, 1–14. https://doi.org/10.1145/3719209 Taherdoost, H., & Madanchian, M. (2023). Artificial intelligence and knowledge management: Impacts, benefits, and implementation. Computers, 12(4), 72. https://doi.org/10.3390/computers12040072 Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18(7), 509–533. https://doi.org/10.1002/(SICI)1097-0266(199708)18:7<509::AID-SMJ882>3.0.CO;2-Z Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367. https://doi.org/10.1177/1534484305278283 Tranfield, D., Denyer, D., & Smart, P. (2003). Towards a methodology for developing evidence‐informed management knowledge by means of systematic review. British Journal of Management, 14(3), 207–222. https://doi.org/10.1111/1467-8551.00375 Vaccaro, M., Almaatouq, A., & Malone, T. (2024). When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour, 8(12), 2293–2303. https://doi.org/10.1038/s41562-024-02024-1 von Krogh, G. (2018). Artificial intelligence in organizations: New opportunities for phenomenon-based theorizing. Academy of Management Discoveries, 4(4), 404–409. https://doi.org/10.5465/amd.2018.0084 Wach, K., Duong, C. D., Ejdys, J., Kazlauskaitė, R., Korzynski, P., Mazurek, G., Paliszkiewicz, J., & Ziemba, E. (2023). The dark side of generative artificial intelligence: A critical analysis of controversies and risks of ChatGPT. Entrepreneurial Business and Economics Review, 11(2), 7–30. https://doi.org/10.15678/EBER.2023.110201 Walsh, J. P., & Ungson, G. R. (1991). Organizational memory. Academy of Management Review, 16(1), 57–91. https://doi.org/10.5465/amr.1991.4278992 Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Xu, Z., Jain, S., & Kankanhalli, M. (2024). Hallucination is inevitable: An innate limitation of large language models. arXiv preprint. https://doi.org/10.48550/arXiv.2401.11817 Zack, M. H. (1999). Developing a knowledge strategy. California Management Review, 41(3), 125–145. https://doi.org/10.2307/41166000

Author: Jose Garcia Affiliation: Swiss International University (SIU) ORCID ID: 0009-0001-2055-9608 Submitted 02 May 2026; Revised 20 June 2026; Revised 27 July 2026; Accepted 4 August 2026; Available online 11 August 2026; Version of Record 11 August 2026. Doi: https://doi.org/10.65326/u7y10029 Volume 3, December 2026, (10029) Abstract Two decades of guidance on equitable international partnership have coincided with little movement in the distribution of authorship, funding custody and agenda control across higher education collaborations. This article explains that persistence by separating two classes of governance instrument. Procedural instruments — memoranda, steering committees, consultation rounds, codes of conduct, equity checklists, reflexivity statements — regulate voice, visibility and process. Allocative instruments transfer custody of one of the four goods a partnership actually distributes: agenda authority, resource custody, epistemic authority, and the direction in which accountability runs. The article develops this distinction into a framework and uses it to specify a condition it calls participatory settlement, a stable state in which procedural instruments raise satisfaction with process and perceived legitimacy while leaving all four goods where they were, and in which the raised legitimacy then reduces pressure for allocative change. The account reconciles two literatures that have talked past each other, one treating partnership failure as a problem of design and the other treating design as an instrument of legitimation. Five propositions specify when a procedural instrument converts into an allocative one: when allocation is the default rather than the request, when a party outside the partnership holds enforcement standing, when transfers cannot be reversed within a funding cycle, when agenda authority and resource custody move together, and when evaluation measures allocation rather than process. The framework is conceptual, derived from published evidence, and offered for empirical test. Keywords: partnership governance, internationalisation of higher education, capacity building, epistemic justice, research collaboration, accountability, SDG 17 1. Introduction Across 7,100 health research articles about sub-Saharan Africa published between 2014 and 2016, authors based in the country the paper studied held 52.9% of first-author positions. Where the collaboration included a partner at a top United States university, that share fell to 23.0%, and 13.5% of the papers carried no author from the country at all (Hedt-Gauthier et al., 2019). A later study of one long-running Southern African consortium found almost none of the outright exclusion the term parachute research describes, and still reported that researchers from low- and lower-middle-income countries were roughly a fifth as likely as high-income-country colleagues to hold either the first or the last position on the byline (Skrivankova et al., 2023). Medical education journals show the same shape from a different angle: five countries supply first or last authors on 70% of papers, and 43% of the world's countries appear on none (Wondimagegn et al., 2023). The figures matter beyond the byline because of what has been loaded onto partnership in the intervening years. Higher education institutions are now written into the development agenda as producers of the research, the graduates and the policy advice on which the Sustainable Development Goals depend, and partnership is the form through which that contribution is meant to travel between systems (Chankseliani & McCowan, 2021). Authorship is the visible trace of a distribution that runs through every part of such an arrangement. Abimbola (2019) named the underlying problem precisely: scholarship about a place is written for an audience somewhere else, and the resulting work carries the marks of the audience rather than the site. This is the record of a period during which institutional effort to make partnership fairer was unusually concentrated. A global code of conduct for research in resource-poor settings was drafted and disseminated (Schroeder et al., 2019). A twenty-item equity checklist for bilateral partnerships was published and taken up (Hodson et al., 2023). Journal editors agreed a consensus statement recommending that manuscripts from international partnerships carry structured reflexivity statements, published alongside the accepted paper (Morton et al., 2022). Funders adopted the language of equitable partnership in their calls, and researchers began to interrogate what that language does and does not reach (Charani et al., 2022; Flint et al., 2022). The instruments multiplied. The distributions barely moved. The obvious explanations do not survive much pressure. Implementation lag would predict a slow but visible trend, and the authorship series shows drift rather than direction. Insufficient uptake would predict that partnerships adopting the instruments diverge from those that do not, which is not what the consortium studies report: collaborations with mature governance arrangements, joint committees and explicit equity commitments produce byline distributions that look much like everyone else's (Skrivankova et al., 2023). Something in the relation between the instruments and the outcomes is misdescribed. I argue that the instruments and the outcomes are largely independent, and that the reason is analytic rather than motivational. Most partnership governance instruments regulate process — who is consulted, who sits on which committee, what is disclosed, what is declared. The outcomes at issue are matters of allocation — who controls the research question, who holds the grant, whose knowledge is treated as evidence, and to whom the partnership owes an account of itself. A partnership can improve every process measure without transferring custody of any of those things. When it does, the result is not hypocrisy. It is a stable arrangement in which the appearance of shared governance is genuine at the level of participation and absent at the level of allocation, and in which the credibility earned by participation removes the pressure that allocation would otherwise face. I call this condition participatory settlement. The contribution is a framework that distinguishes governance instruments by what they allocate, an account of why the two classes are so easily confused, and five propositions specifying the conditions under which a procedural instrument converts into an allocative one. The framework speaks to two literatures that have addressed the same object without meeting. One treats partnership difficulty as a design problem soluble by better platforms, clearer contracts and stronger deliberation (Ansell & Gash, 2018; Dentoni et al., 2018; Pattberg & Widerberg, 2016). The other treats governance design as part of the problem, arguing that the instruments encode the asymmetries they claim to correct (Kumar et al., 2024; Mormina & Istratii, 2021; Ndlovu-Gatsheni, 2021). Neither has specified which instruments do which. That specification is what the framework provides. The article sets out the two accounts and the tension between them, identifies the four goods a partnership allocates, classifies governance instruments by whether they move those goods, develops participatory settlement as the condition that follows when procedural instruments are used alone, states five propositions on conversion, and closes with implications for theory, for institutions and funders, and for research, together with the limits of the account. 1.1. Analytical Approach This is a conceptual article. It advances no empirical claim of its own, reports no primary data, and derives its framework from published work whose findings are attributed to their sources throughout. Because a conceptual argument can be checked only if its evidential base is visible, the procedure by which that base was assembled is set out here. Sources were identified through structured searches of OpenAlex and PubMed, supplemented by targeted retrieval from publisher sites and citation chaining on the most central retrieved papers. Fourteen search strings were run across five construct families: partnership and network governance (search terms combining collaborative governance, cross-sector partnership, multi-stakeholder, accountability, legitimacy); internationalisation of higher education (internationalisation, transnational education, partnership, governance); research capacity development (research capacity strengthening, North-South partnership, equity, asymmetry); knowledge exchange and the university's external missions (quadruple helix, university-industry, knowledge transfer, third mission); and the infrastructure of scholarly communication (authorship inequality, article processing charges, language, open access). The window was 2014 to 2026, extended backwards where a work was needed to establish a position that later literature argues with. Three criteria governed inclusion. A source had to be peer reviewed or a scholarly monograph; it had to bear on the allocation of authority, resources, knowledge claims or accountability within collaborative arrangements, rather than on partnership outcomes alone; and its metadata had to be confirmable against an indexed record. Sixty-four sources met all three. Each was logged with full bibliographic detail, the verification path, and a one-line statement of what the source claims, so that every proposition below can be traced to the evidence supporting it. Sources that could not be confirmed were discarded rather than reconstructed. The framework was derived in three steps. First, recurrent objects of dispute were extracted from the empirical partnership literature, meaning what partners reported fighting over or reported being unable to influence. These clustered into four goods. Second, the governance instruments described across the same literature were sorted by whether their operation changes custody of any of those goods. Third, the cases in which instruments did appear to move custody were compared with those in which they did not, and the differences between them were stated as conditions. The propositions are the output of that third step. Two limits on the procedure should be recorded. The evidence base is uneven by field, since health research partnerships are documented far more thoroughly than partnerships in engineering, the humanities or professional education. The literature is also predominantly English-language, which is itself one of the asymmetries the article analyses (Ramírez-Castañeda, 2020). Widening the construct families beyond health and drawing the governance apparatus from public administration rather than from higher education alone reduces neither limit to zero. 2. Two Accounts of What Partnership Governance Does The first account treats governance as engineering. Its object is the collaborative arrangement itself, and its question is what design makes the arrangement work. Ansell and Gash (2018) describe collaborative platforms as organisations holding dedicated competences and resources for hosting repeated collaborations, and identify strategic intermediation and design rules as what makes them adapt and survive. Dentoni et al. (2018) match three governance processes — deliberation, decision-making and enforcement — to three dimensions along which problems are difficult, and argue that partnerships fail when the process is mismatched to the problem. Pattberg and Widerberg (2016) distil nine conditions for partnership performance from a decade of scholarship. In the higher education literature the same logic appears in work on partnership sustainability, which identifies threats to long-term collaboration and the conditions that offset them (Lanford, 2019), in the analysis of transnational teaching networks, where governance form and network size predict what kinds of innovation a consortium attempts (Fumasoli & Rossi, 2021), and in the account of higher education steering as a multi-level arrangement in which supranational, national and institutional authority are distributed and stakeholder involvement partly serves to legitimate policy (Fumasoli, 2015). This account has produced real knowledge. It has established that contracts and relationships are not substitutes: contractual governance shapes relational norms and partner contribution rather than displacing trust (Benítez-Ávila et al., 2018). It has shown that partnerships survive incompatible institutional logics when resource interdependence is lower, and break under the combination of incompatible logics and high dependence (Ashraf et al., 2017). It has engaged asymmetry directly rather than assuming it away: Ran and Qi (2018) set out six contingency factors governing power sharing and conclude that designing to contingencies is more productive than pursuing balanced power for its own sake. And it has been honest about its own limits. Howlett and Ramesh (2014) argued a decade ago that network and collaborative modes fail in their own characteristic ways, and that the claim they combine the virtues of hierarchy and market rests on faith rather than evidence. The second account treats governance as legitimation. Its object is the relation between the partners, and its question is what the arrangement does to an asymmetry that predates it. Abimbola et al. (2021) set out the forms that asymmetry takes across global collaboration and argue that it is structural rather than attitudinal, which means it is not addressed by improving the conduct of the parties. Mormina and Istratii (2021) argue that research capacity development remains technocratic and unreflexive about its own normative direction, so that its instruments carry an unexamined account of what capacity is for and who decides. Flint et al. (2022) trace how United Kingdom funding models place the grant, the compliance burden and the agenda on the northern side, so that equity language does not reach the terms on which the partnership operates. Kumar et al. (2024) state the point about instruments directly: existing guidelines and checklists do not touch decision-making power, resource distribution or funder accountability. Ndlovu-Gatsheni (2021) locates the asymmetry further back, in a division of intellectual labour that treats the position from which knowledge is produced as though it were nowhere in particular. The parallel argument about the university's internal arrangements holds that curricular and epistemic change requires alteration to who governs, and that additions to reading lists do not substitute for it (Bhambra et al., 2018). Read across a whole regional policy field, education agendas carry the same difficulty: Tikly (2019) treats African education-for-sustainable-development policy as an artefact of the postcolonial condition, in which the reforming instrument reproduces the relation it addresses. Interview evidence from partners on the receiving side of capacity building supports this account in a specific way. Zambian health researchers attributed publication disputes, agenda dictation and conflicts over sample and data ownership to the concentration of funding on the northern side of the collaboration, rather than to any deficiency in the governance forms in use (Matenga et al., 2019). Members of a North-South-South network reported that donor agendas set the topics and that dependence on external calls left little room for locally initiated work; South-South collaboration within the network, the arrangement's stated purpose, remained minimal (Van der Veken et al., 2017). Angolan and Mozambican academics in a European-led project valued the training and reported that the project had not engaged their institutional context, and that the asymmetry the project was designed to reduce structured the project itself (Craveiro et al., 2020). At a Ghanaian elite university, researchers described a dependency produced jointly by external funding structures and internal institutional weakness, which no partnership agreement addressed (Nkansah et al., 2025). The two accounts are not simply opposed, and the assumption that they are has cost the field a decade. The engineering account is right that instruments matter and that partnerships with better arrangements do better on the things those arrangements govern. The legitimation account is right that partnerships with better arrangements have not produced different distributions of authorship, funding or agenda control. Both can hold if the instruments in question govern something other than those distributions. That is the possibility neither literature has examined, because neither has asked what a partnership instrument allocates. A third position points toward the answer without reaching it. Fougère and Solitander (2020) argue that consensus-oriented multi-stakeholder initiatives cannot be democratic on their own terms, because adversarial relations are displaced outside the initiative rather than worked through inside it. Chambers et al. (2022), analysing 32 co-production initiatives, found that effort was spent closing down disagreement between competing agendas rather than navigating it. Morrison et al. (2019) show that power in polycentric arrangements operates through rule setting, issue construction and implementation, and that the polycentricity literature has treated it as a residual. West et al. (2019) diagnose a related failure inside co-production itself, which they argue remains captive to a linear assumption that knowledge is produced first and applied afterwards, so that collaborative form does not disturb the sequence. Staffa et al. (2022) propose dissenting-within as a mode of collaborative knowledge production precisely because agreement-seeking arrangements have no place to put disagreement. Each of these observes that something is being managed rather than settled. None identifies what. 3. What Partnerships Allocate Read across the empirical literature, partners in international higher education collaborations report contesting, or being unable to influence, four things. They are distinguishable in principle and separable in practice, which is what makes them useful as a set. Agenda authority is control over what the collaboration is about: the research question, the curricular content, the definition of the problem the partnership addresses. Network members report losing it not through exclusion from meetings but through the structure of the funding call that preceded the meeting (Van der Veken et al., 2017). Where a partnership responds to a call written elsewhere, the agenda arrives with the money and the joint committee's task is implementation. Academics in low-income East African systems describe the resulting arrangement as non-reciprocal, and place it alongside weak internationalisation structures and brain drain among the constraints they face (Moshtari & Safarpour, 2024). Mormina and Istratii (2021) make the sharper version of the point: capacity development that does not ask capacity for what has answered the question by default. Resource custody is control over money, equipment and data, meaning not the volume received but the position in which it is held. The distinction matters because sub-award status and co-investigator status differ in almost every respect that governs a partnership's operation: who reports to the funder, who bears compliance cost, who can reallocate a budget line mid-project, who may terminate. Flint et al. (2022) show how United Kingdom funding architecture assigns those positions before any partnership agreement is written. Charani et al. (2022) identify funders as the actors whose practices set the terms, and call for direct funding to institutions rather than through an intermediary in a high-income country. Epistemic authority is standing to determine what counts as evidence and as a contribution: which methods are recognised, which language the work is written and assessed in, which venues confer credit. It is the good most often described and least often treated as allocable. Ndlovu-Gatsheni (2021) frames it as the position from which knowledge is enunciated. Cummings et al. (2023) extend Fricker's epistemic injustice to its structural and systemic forms, including linguistic injustice. Ramírez-Castañeda (2020) supplies the concrete measure: among Colombian doctoral students in the biological sciences, 43.5% reported an article rejected or returned on grounds of English grammar, and 33% declined to attend international meetings because presentation was required in English. Morley et al. (2018) report epistemic exclusion as a persistent feature of academic mobility even where the material gains are substantial. Accountability direction settles a matter partnerships rarely put explicitly, which is to whom the arrangement owes an account of itself. Where reporting runs upward to a funder, the partnership's performance is defined by the funder's categories, and the partner institution's own priorities enter only where they coincide. Mason (2020) reviewed the evidence on disclosure in sustainability governance and found that transparency improves the communication of affected parties' interests while doing little to give them capacity to sanction or steer. Nonet et al. (2022) name the same problem as the material-coupling failure of multi-stakeholder engagement, distinct from relational and cognitive alignment. In development cooperation the debate about ownership has circled this point for two decades without settling it (Morvaridi & Hughes, 2018). Two features of this set matter for what follows. The goods are separable: a partnership can devolve resource custody while retaining agenda authority, and the practice of sub-granting an implementation budget for a question defined elsewhere is common enough to have its own critique (Franzen et al., 2017). And they are jointly sufficient to account for the reported grievances. Every complaint in the interview studies cited above resolves into a claim about the custody of one of the four, which is the empirical warrant for treating the set as complete for present purposes rather than merely convenient. 4. Procedural and Allocative Instruments The framework's central move follows directly. Sort the governance instruments used in international higher education partnerships by whether their operation transfers custody of any of the four goods. Two classes appear, and the line between them does not track the line between weak and strong instruments, or between soft and hard ones. Procedural instruments regulate participation, deliberation, disclosure and declaration. They determine who is present, what is discussed, what is made visible and what is stated on the record. Memoranda of understanding, joint steering committees, consultation rounds, partnership principles, codes of conduct, equity checklists and reflexivity statements are all of this kind. So, less obviously, are most capacity-building activities: training transfers skill to individuals without changing who decides what those skills are used for. Allocative instruments transfer custody. Direct grant-holding by the partner institution moves resource custody. A partner-defined call, or a requirement that the research question originate with the partner, moves agenda authority. First-author defaults tied to the site of data generation, publication in partner-country venues, funded translation and multilingual review move epistemic authority. Reporting obligations that run to the partner institution and to affected communities, rather than only to the funder, move accountability direction. Table 1 sorts the instruments the literature describes. The pattern in the final column is the finding: the instruments that have proliferated fastest are almost entirely procedural. Table 1 Governance instruments in international higher education partnerships, classified by the goods whose custody they move Instrument Class AG RC EA AD Documented in Memorandum of understanding; joint steering committee Procedural – – – – Lanford (2019); Fumasoli & Rossi (2021) Partnership principles and codes of conduct Procedural – – – – Schroeder et al. (2019) Equity checklists applied across the project cycle Procedural – – – – Hodson et al. (2023) Structured reflexivity statements published with the article Procedural – – ○ – Morton et al. (2022) Training, mentorship and fellowship provision Procedural – – ○ – Franzen et al. (2017); Kasprowicz et al. (2020) Consultation on an agenda already funded Procedural – – – – Van der Veken et al. (2017) Direct grant-holding by the partner institution Allocative – ● – ● Charani et al. (2022) Partner-originated calls and question-setting Allocative ● ○ – – Mormina & Istratii (2021) First-author default tied to site of data generation Allocative – – ● – Hedt-Gauthier et al. (2019); Skrivankova et al. (2023) Funded translation; multilingual review and dissemination Allocative – ○ ● – Ramírez-Castañeda (2020); Cummings et al. (2023) Fee waivers and diamond open access for partner authors Allocative – ○ ● – Borrego (2023); Butler et al. (2023) Reporting obligations owed to partner institutions Allocative – – – ● Mason (2020); Nonet et al. (2022) Consortium leadership vested in partner-country institutions Allocative ● ● ○ ● Kasprowicz et al. (2020) Note. AG = agenda authority; RC = resource custody; EA = epistemic authority; AD = accountability direction. ● indicates that the instrument transfers custody of the good; ○ indicates a partial or conditional transfer; – indicates no transfer. Classification is analytic: an instrument's class follows from what its operation moves, not from its formal strength. Compiled from the sources listed in the final column. Four qualifications keep the distinction from hardening into a dismissal. First, procedural instruments are preconditions, not decoration. Without a forum, an agreement and a record, there is no place in which allocation can be raised, contested or even named; the interview studies show partners identifying asymmetries precisely because a partnership existed in which they could be identified (Craveiro et al., 2020). Second, several instruments sit on the boundary. Reflexivity statements move a small quantity of epistemic authority by making the division of labour assessable by editors and readers, and Morton et al. (2022) designed them to do exactly that. Third, the class of an instrument is a property of its operation rather than its text. A code of conduct that a funder makes a condition of disbursement operates allocatively even though its clauses are procedural, which is the first of the conversion conditions developed below. A fourth qualification cuts the other way and should be stated by its strongest advocates. Caniglia et al. (2023) argue that the quality of collaborative knowledge production turns on the situated judgement of the people doing it rather than on compliance with any procedure, which implies that classifying instruments by what they allocate may miss what actually determines whether a partnership goes well. The framework does not dispute that. It claims only that judgement operates on whatever the arrangement has placed within reach, and that custody determines the reach. 5. Participatory Settlement If procedural and allocative instruments were merely different, the pattern in Table 1 would be an oversight and the remedy would be obvious. The framework's second claim is that the two classes interact, and that the interaction is what makes the pattern stable. Procedural instruments do something that matters to the parties who hold the four goods. They generate legitimacy. Participation, disclosure and declaration produce evidence that the arrangement is fair, and that evidence circulates: to funders assessing a renewal, to institutional leaders reporting on internationalisation, to editors assessing a submission, to the partners themselves. The finding that trust, confidence and perceived equity drive the legitimacy of governance arrangements (Turner et al., 2016) holds here in a form that cuts against equity, because the perception can be produced by process alone. Legitimacy so generated is not counterfeit. Consultation did occur, the committee did meet, the statement was filed. The consequence is that legitimacy and allocation come apart, and that the first substitutes for the second in the political economy of the partnership. A collaboration displaying good process faces less pressure to change custody than one displaying poor process, because the visible indicators of fairness are process indicators. Funders assess partnership quality through governance documentation. Institutions report internationalisation through partnership counts and agreements (Alexiadou & Rönnberg, 2022; Tight, 2021). Journals, until very recently, assessed nothing about the division of labour at all. Where every monitoring channel reads process, process is what improves. This is participatory settlement: a stable configuration in which procedural instruments accumulate, satisfaction with process rises, perceived legitimacy rises with it, and custody of agenda authority, resource custody, epistemic authority and accountability direction remains where it began — with the pressure for allocative change falling as legitimacy rises. Figure 1 sets out the framework and the reinforcing loop that holds the configuration in place. Figure 1 The allocation framework: procedural and allocative pathways in partnership governance Note. The upper pathway shows procedural instruments producing participation, deliberation and disclosure, which generate perceived legitimacy. Dashed lines mark the reinforcing loop: legitimacy reduces the pressure for allocative change, and reduced pressure lowers the rate at which allocative instruments are adopted. The lower pathway shows allocative instruments transferring custody of the four goods. Conditions C1 to C5 are the conversion conditions stated as Propositions 1 to 5. The configuration resolves several observations that are otherwise puzzling. Partnerships with mature governance and explicit equity commitments produce authorship distributions indistinguishable from those without them (Skrivankova et al., 2023) because the commitments operate on a channel that does not reach the byline. Partners report genuine benefit and persistent asymmetry in the same interview, a pattern so consistent across the capacity-building studies that it reads as contradiction until the two are recognised as reports about different goods (Craveiro et al., 2020; Matenga et al., 2019). And the agonistic critique's central observation follows without recourse to bad faith: consensus-oriented arrangements suppress dissent (Fougère & Solitander, 2020) because dissent is the only route by which an allocative question reaches a body constituted to handle procedural ones. A finding from outside higher education fits the same shape. Mason (2020) found that disclosure improves the articulation of affected parties' interests while doing little to enable them to sanction or steer. That is participatory settlement stated in the vocabulary of environmental governance, and its appearance in a separate domain suggests the configuration is a property of governance arrangements under certain conditions rather than a feature of any one field's practice. Participatory settlement is not a claim that the parties holding the four goods act strategically to retain them. The configuration requires no such intention. It follows from two ordinary facts: procedural instruments are cheaper, faster and less contested to adopt than allocative ones, and the channels through which partnerships are monitored read process. Where those two hold, the configuration forms whatever anyone intends. That is what separates this account from the stronger versions of the legitimation critique, which read the instruments as capture (Meagher, 2021). Capture is one route into the configuration, not the only one, and the account does not depend on it. 6. When Procedure Becomes Allocation: Five Propositions The framework would be of little use if procedural instruments never converted. They do. The cases in which the literature reports movement share features that the cases without movement lack, and those features can be stated as conditions. The propositions below are offered for empirical test; each specifies an observable that would count against it. Proposition 1 (default rule). A procedural instrument converts to allocative operation when it establishes allocation as the default state rather than as an outcome available on request. An instrument that permits a partner institution to hold the grant leaves resource custody where the transaction cost of asking places it. An instrument that requires the partner institution to hold the grant unless a documented reason exists moves it. The waiver literature supplies the test case: fee waivers for authors from low-income settings are available on application and, reviewed across two decades of evidence, have not extended access as intended, in part because the criteria for qualifying and the act of applying carry their own costs (Borrego, 2023). Diamond open access, which requires no application, allocates by default. Proposition 2 (external standing). A procedural instrument converts when a party outside the partnership holds standing to enforce it. Instruments enforced within a partnership are enforced by the parties who hold the goods, which makes enforcement conditional on the outcome it is meant to produce. The journal-side interventions are the clearest instance: the consensus statement on equitable authorship works through editors, who are outside the collaboration, control something the collaboration needs, and assess the reflexivity statement as part of the decision to publish (Morton et al., 2022). Funders occupy the equivalent position on the resource side, which is why Charani et al. (2022) direct their argument to them rather than to the partners. Proposition 3 (within-cycle irreversibility). A transfer of custody holds when it cannot be reversed within the funding cycle in which it occurs. Devolved authority that can be recalled is delegated authority, and delegation does not change custody. A budget line released at a project's start and administered by the partner institution for the project's duration allocates; the same sum released tranche by tranche against milestones assessed by the lead does not. This is why the sustainability of partnership arrangements is a question about custody rather than about goodwill (Lanford, 2019), and why consortium models that vest leadership in partner-country institutions for the full award period behave differently from those that rotate a chair (Kasprowicz et al., 2020). Proposition 4 (bundling). Agenda authority and resource custody must move together or neither transfer holds. Devolving money for a question defined elsewhere converts the partner into a contractor, a form the capacity development literature has described critically for two decades (Franzen et al., 2017). Devolving the question without the money produces an agenda the partner cannot pursue, which is the position the network members in Van der Veken et al. (2017) described. The prediction is specific and falsifiable: partnerships transferring one without the other should show no improvement over partnerships transferring neither, on any measure of the third and fourth goods. Proposition 5 (measurement of allocation). Where evaluation measures allocation rather than process, procedural instruments convert; where it measures process, they do not. The reinforcing loop in Figure 1 runs through monitoring, and monitoring is the point at which it can be cut. Bibliometric studies of authorship position have already demonstrated that allocation is measurable at scale (Hedt-Gauthier et al., 2019; Wondimagegn et al., 2023), and the caution that affiliation data alone cannot distinguish extractive from collaborative practice is well taken (Purnell, 2024). Grant-holding position, agenda origin, publication language and reporting direction are all recordable at the partnership level, and none currently appears in the reporting frameworks through which internationalisation is assessed. 7. Discussion 7.1. Theoretical Implications The framework closes the gap between the engineering and legitimation accounts by locating their disagreement in an unexamined premise both share: that partnership instruments operate on a single dimension along which arrangements are better or worse. Once instruments are sorted by what they allocate, both accounts turn out to be locally correct. Design does determine how well the deliberative and coordinating work of a partnership is done, which is what the collaborative governance literature has measured (Ansell & Gash, 2018; Dentoni et al., 2018). Design has not determined the distribution of authority, resources, epistemic standing or accountability, which is what the critical literature has measured (Kumar et al., 2024; Mormina & Istratii, 2021). The disagreement was about which dimension governance operates on, and the answer is that it operates on both, through different instruments, with different conditions of effectiveness. The account also supplies something the polycentricity and network governance literatures have looked for. Morrison et al. (2019) called power in polycentric arrangements a black box and identified rule setting, issue construction and implementation as the sites at which it acts. The four goods specify what is at stake at each site, and the procedural-allocative distinction specifies which instruments touch them. Carlisle and Gruby's (2019) separation of the definitional attributes of polycentricity from the conditions under which it functions has the same structure as the argument here, and the propositions can be read as conditions of that kind. That the configuration appears in environmental disclosure (Mason, 2020) and in multi-stakeholder engagement for the development goals (Nonet et al., 2022) suggests it is not specific to higher education, though this article does not test that. For research on higher education specifically, the framework reframes internationalisation as an allocation question. The literature has documented that national internationalisation strategies serve domestic priorities (Alexiadou & Rönnberg, 2022; Tight, 2021), that market-making actors reinforce existing asymmetries in the global south (Robertson & Komljenović, 2016), and that institutions use mobility to hold together conflicting mandates without structural change (Majee & Ress, 2020; Stein et al., 2016). Each of these is a claim about what internationalisation allocates. The critical scholarship on the ethics of internationalisation has argued that remedial efforts carry the assumptions that produced the problem (Pashby & Andreotti, 2016; Stein, 2016). Participatory settlement gives that argument a specifiable form: the remedial instruments carry the assumptions because they operate on process, and process is the dimension on which the asymmetry does not sit. The same reframing bears on how the sector's contribution is assessed. Findler et al. (2019) found that the literature on what higher education institutions do for sustainable development is dominated by project-level case studies and short of system-level analysis, which is what one would expect where the reporting channels record activity rather than allocation. 7.2. Implications for Institutions, Funders and Journals The practical reading of the framework is a sequencing argument rather than a rejection of current practice. Institutions entering a partnership should treat the four goods as a design checklist and record their custody at the outset, in the same document that establishes the committee structure. Where custody is not transferred, the partnership can still be worth forming; what the framework rules out is describing it as equitable on procedural evidence. Funders sit where Propositions 1 to 3 can be satisfied at once. Direct grant-holding by partner institutions moves resource custody and accountability direction together; making it the default rather than an option satisfies the first proposition; and funders enforce from outside the partnership, satisfying the second. The proposal is not new (Charani et al., 2022), and the framework explains why it has more effect than the checklist approaches with which it competes for attention. Journals and publishers control the epistemic good more completely than any other actor, and the routes by which they do so are already documented. Author-side charges exclude by price and waivers have not compensated for it (Borrego, 2023; Butler et al., 2023; Frank et al., 2023). English-only submission and review impose a cost measurable in rejections and non-attendance (Ramírez-Castañeda, 2020). Reflexivity statements assessed by editors satisfy the external-standing condition and are already in use (Morton et al., 2022). Funded translation and diamond open access satisfy the default condition; waiver schemes do not. The framework is silent on one thing that matters to practice. It specifies the conditions under which custody moves; it does not specify whether moving custody improves the substantive quality of the work, and the evidence on that question is thin in every direction. Consortium models that vest leadership in partner-country institutions report benefits (Kasprowicz et al., 2020), and the argument for them in that literature is made on grounds of equity and sustainability rather than of output quality. The framework should not be read as promising a quality gain it does not establish. 7.3. Limitations Four limits bound the account. It is conceptual: the four goods, the two instrument classes and the settlement condition are derived from published evidence and have not been operationalised or tested, and the propositions are offered as testable rather than as tested. The evidence base is uneven, drawn disproportionately from health research partnerships, whose funding architecture is unusually concentrated and whose authorship conventions are unusually well studied; whether the configuration holds where funding is more diffuse is unknown. The four goods are claimed to be sufficient for the grievances the literature reports, which is weaker than a claim to exhaustiveness, and a further good such as control over infrastructure or over the training pipeline might prove separable on evidence not examined here. And participatory settlement is specified as a configuration rather than measured as one; distinguishing it empirically from slow implementation requires longitudinal data on custody that no current reporting framework collects, which is a limitation of the account and an argument for Proposition 5. 7.4. Future Research The work the framework most needs is measurement. Partnership-level indicators of custody for each of the four goods would allow a test of whether custody moves independently of procedural indicators. Grant-holding position, agenda origin, first-author position by site of data generation, publication language and reporting direction are all recordable, and the bibliometric work shows the authorship component is already tractable at scale (Skrivankova et al., 2023; Wondimagegn et al., 2023). Proposition 4 admits a more direct test, comparing partnerships that transfer agenda authority and resource custody together against those transferring one alone; existing funding portfolios contain both arrangements in sufficient numbers. Beyond health and beyond research, transnational teaching consortia, joint degree programmes and university-industry arrangements allocate the same four goods under different funding structures, and the knowledge exchange literature has mapped their instruments without asking what those instruments move (Miller et al., 2018; Skute et al., 2019). Whether the configuration appears where the fourth helix is civil society rather than a partner university is an open and answerable question (Cai & Etzkowitz, 2020; Compagnucci & Spigarelli, 2020). 8. Conclusion The instruments of partnership governance have multiplied while the distributions they were meant to correct have held. This article has argued that the two facts are compatible because most of those instruments regulate process, and the distributions are matters of allocation. Four goods are allocated in an international higher education partnership: authority over the agenda, custody of resources, standing to determine what counts as knowledge, and the direction in which accountability runs. Governance instruments either move custody of these or they do not, and the class that has grown fastest does not. The result is participatory settlement, in which good process generates the legitimacy that would otherwise be the pressure for allocative change. The argument does not counsel abandoning procedural instruments. It counsels sequencing them behind allocative ones and reading them for what they are. It also yields five conditions under which procedure converts: allocation by default, enforcement from outside, transfers that cannot be recalled within the cycle, agenda and resources moving together, and evaluation that measures where the four goods sit. Each is stated so that evidence could contradict it. The framework's usefulness will be settled by whether anyone collects that evidence, and the case for collecting it is that a field which has spent twenty years improving what it measures has not yet measured what it says it wants to change. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Use of AI and AI-assisted technologies. Generative artificial intelligence tools were used solely to improve the readability and language of the manuscript. The author conceived the research question, developed the analytical framework, appraised and interpreted all sources, and drew all conclusions. The author reviewed and edited all content and take full responsibility for the accuracy and integrity of the published work. Ethics Statement. This article is a conceptual analysis of published scholarly literature. It involved no human participants, no animal subjects and no primary data collection, and ethical approval was therefore not required. Data Availability. No datasets were generated or analysed. All sources supporting the argument are cited in the reference list and are publicly available through the identifiers given there. References Abimbola, S. (2019). The foreign gaze: Authorship in academic global health. BMJ Global Health, 4(5), e002068. https://doi.org/10.1136/bmjgh-2019-002068 Abimbola, S., Asthana, S., Montenegro, C., Guinto, R. R., Jumbam, D. T., Louskieter, L., Kabubei, K. M., Munshi, S., Muraya, K., Okumu, F., Saha, S., Saluja, D., & Pai, M. (2021). Addressing power asymmetries in global health: Imperatives in the wake of the COVID-19 pandemic. PLoS Medicine, 18(4), e1003604. https://doi.org/10.1371/journal.pmed.1003604 Alexiadou, N., & Rönnberg, L. (2022). Transcending borders in higher education: Internationalisation policies in Sweden. European Educational Research Journal, 21(3), 504-519. https://doi.org/10.1177/1474904120988383 Ansell, C., & Gash, A. (2018). Collaborative platforms as a governance strategy. Journal of Public Administration Research and Theory, 28(1), 16-32. https://doi.org/10.1093/jopart/mux030 Ashraf, N., Ahmadsimab, A., & Pinkse, J. (2017). From animosity to affinity: The interplay of competing logics and interdependence in cross-sector partnerships. Journal of Management Studies, 54(6), 793-822. https://doi.org/10.1111/joms.12273 Benítez-Ávila, C., Hartmann, A., Dewulf, G., & Henseler, J. (2018). Interplay of relational and contractual governance in public-private partnerships: The mediating role of relational norms, trust and partners' contribution. International Journal of Project Management, 36(3), 429-443. https://doi.org/10.1016/j.ijproman.2017.12.005 Bhambra, G. K., Gebrial, D., & Nişancıoğlu, K. (Eds.). (2018). Decolonising the university. Pluto Press. https://doi.org/10.2307/j.ctv4ncntg Borrego, A. (2023). Article processing charges for open access journal publishing: A review. Learned Publishing, 36(3), 359-378. https://doi.org/10.1002/leap.1558 Butler, L.-A., Matthias, L., Simard, M.-A., Mongeon, P., & Haustein, S. (2023). The oligopoly's shift to open access: How the big five academic publishers profit from article processing charges. Quantitative Science Studies, 4(4), 778-799. https://doi.org/10.1162/qss_a_00272 Cai, Y., & Etzkowitz, H. (2020). Theorizing the Triple Helix model: Past, present, and future. Triple Helix Journal, 7(2-3), 189-226. https://doi.org/10.1163/21971927-bja10003 Caniglia, G., Freeth, R., Luederitz, C., Leventon, J., West, S., John, B., Peukert, D., Lang, D. J., von Wehrden, H., Martín-López, B., Fazey, I., Russo, F., von Wirth, T., Schlüter, M., & Vogel, C. (2023). Practical wisdom and virtue ethics for knowledge co-production in sustainability science. Nature Sustainability, 6(5), 493-501. https://doi.org/10.1038/s41893-022-01040-1 Carlisle, K. M., & Gruby, R. L. (2019). Polycentric systems of governance: A theoretical model for the commons. Policy Studies Journal, 47(4), 927-952. https://doi.org/10.1111/psj.12212 Chambers, J. M., Wyborn, C., Klenk, N. L., Ryan, M., Șerban, A., Bennett, N. J., Brennan, R., Charli-Joseph, L., Fernández-Giménez, M. E., Galvin, K. A., Goldstein, B. E., Haller, T., Hill, R., Múnera-Roldán, C., Nel, J. L., Österblom, H., Reid, R. S., Riechers, M., Spierenburg, M., & Tengö, M. (2022). Co-productive agility and four collaborative pathways to sustainability transformations. Global Environmental Change, 72, 102422. https://doi.org/10.1016/j.gloenvcha.2021.102422 Chankseliani, M., & McCowan, T. (2021). Higher education and the Sustainable Development Goals. Higher Education, 81(1), 1-8. https://doi.org/10.1007/s10734-020-00652-w Charani, E., Abimbola, S., Pai, M., Adeyi, O., Mendelson, M., Laxminarayan, R., & Rasheed, M. A. (2022). Funders: The missing link in equitable global health research? PLOS Global Public Health, 2(6), e0000583. https://doi.org/10.1371/journal.pgph.0000583 Compagnucci, L., & Spigarelli, F. (2020). The Third Mission of the university: A systematic literature review on potentials and constraints. Technological Forecasting and Social Change, 161, 120284. https://doi.org/10.1016/j.techfore.2020.120284 Craveiro, I., Carvalho, A., & Ferrinho, P. (2020). "Get us partnerships!" - A qualitative study of Angolan and Mozambican health academics' experiences with North/South partnerships. Globalization and Health, 16, Article 33. https://doi.org/10.1186/s12992-020-00562-7 Cummings, S., Dhewa, C., Kemboi, G., & Young, S. (2023). Doing epistemic justice in sustainable development: Applying the philosophical concept of epistemic injustice to the real world. Sustainable Development, 31(3), 1965-1977. https://doi.org/10.1002/sd.2497 Dentoni, D., Bitzer, V., & Schouten, G. (2018). Harnessing wicked problems in multi-stakeholder partnerships. Journal of Business Ethics, 150(2), 333-356. https://doi.org/10.1007/s10551-018-3858-6 Findler, F., Schönherr, N., Lozano, R., Reider, D., & Martinuzzi, A. (2019). The impacts of higher education institutions on sustainable development. International Journal of Sustainability in Higher Education, 20(1), 23-38. https://doi.org/10.1108/ijshe-07-2017-0114 Flint, A., Howard, G., Baidya, M., Wondim, T., Poudel, M., Nijhawan, A., Mulugeta, Y., & Sharma, S. (2022). Equity in Global North-South research partnerships: Interrogating UK funding models. Global Social Challenges Journal, 1(1), 76-93. https://doi.org/10.1332/vqil8302 Fougère, M., & Solitander, N. (2020). Dissent in consensusland: An agonistic problematization of multi-stakeholder governance. Journal of Business Ethics, 164(4), 683-699. https://doi.org/10.1007/s10551-019-04398-z Frank, J., Foster, R., & Pagliari, C. (2023). Open access publishing - Noble intention, flawed reality. Social Science & Medicine, 317, 115592. https://doi.org/10.1016/j.socscimed.2022.115592 Franzen, S. R. P., Chandler, C., & Lang, T. (2017). Health research capacity development in low and middle income countries: Reality or rhetoric? A systematic meta-narrative review of the qualitative literature. BMJ Open, 7(1), e012332. https://doi.org/10.1136/bmjopen-2016-012332 Fumasoli, T. (2015). Multi-level governance in higher education research. In J. Huisman, H. de Boer, D. D. Dill, & M. Souto-Otero (Eds.), The Palgrave international handbook of higher education policy and governance (pp. 76-94). Palgrave Macmillan. https://doi.org/10.1007/978-1-137-45617-5_5 Fumasoli, T., & Rossi, F. (2021). The role of higher education institutions in transnational networks for teaching and learning innovation: The case of the Erasmus+ programme. European Journal of Education, 56(2), 200-218. https://doi.org/10.1111/ejed.12454 Hedt-Gauthier, B. L., Jeufack, H. M., Neufeld, N. H., Alem, A., Sauer, S., Odhiambo, J., Boum, Y., Shuchman, M., & Volmink, J. (2019). Stuck in the middle: A systematic review of authorship in collaborative health research in Africa, 2014-2016. BMJ Global Health, 4(5), e001853. https://doi.org/10.1136/bmjgh-2019-001853 Hodson, D. Z., Etoundi, Y. M., Parikh, S., & Boum, Y. (2023). Striving towards true equity in global health: A checklist for bilateral research partnerships. PLOS Global Public Health, 3(1), e0001418. https://doi.org/10.1371/journal.pgph.0001418 Howlett, M., & Ramesh, M. (2014). The two orders of governance failure: Design mismatches and policy capacity issues in modern governance. Policy and Society, 33(4), 317-327. https://doi.org/10.1016/j.polsoc.2014.10.002 Kasprowicz, V. O., Chopera, D., Waddilove, K. D., Brockman, M. A., Gilmour, J., Hunter, E., Kilembe, W., Karita, E., Gaseitsiwe, S., Sanders, E. J., & Ndung'u, T. (2020). African-led health research and capacity building - Is it working? BMC Public Health, 20, Article 1104. https://doi.org/10.1186/s12889-020-08875-3 Kumar, R., Khosla, R., & McCoy, D. (2024). Decolonising global health research: Shifting power for transformative change. PLOS Global Public Health, 4(4), e0003141. https://doi.org/10.1371/journal.pgph.0003141 Lanford, M. (2019). Long-term sustainability in global higher education partnerships. In Successful global collaborations in higher education institutions (pp. 87-93). Springer. https://doi.org/10.1007/978-3-030-25525-1_9 Majee, U. S., & Ress, S. B. (2020). Colonial legacies in internationalisation of higher education: Racial justice and geopolitical redress in South Africa and Brazil. Compare: A Journal of Comparative and International Education, 50(4), 463-481. https://doi.org/10.1080/03057925.2018.1521264 Mason, M. (2020). Transparency, accountability and empowerment in sustainability governance: A conceptual review. Journal of Environmental Policy & Planning, 22(1), 98-111. https://doi.org/10.1080/1523908x.2019.1661231 Matenga, T. F. L., Zulu, J. M., Corbin, J. H., & Mweemba, O. (2019). Contemporary issues in north-south health research partnerships: Perspectives of health research stakeholders in Zambia. Health Research Policy and Systems, 17, Article 7. https://doi.org/10.1186/s12961-018-0409-7 Meagher, K. (2021). Introduction: The politics of open access - Decolonizing research or corporate capture? Development and Change, 52(2), 340-358. https://doi.org/10.1111/dech.12630 Miller, K., McAdam, R., & McAdam, M. (2018). A systematic literature review of university technology transfer from a quadruple helix perspective: Toward a research agenda. R&D Management, 48(1), 7-24. https://doi.org/10.1111/radm.12228 Morley, L., Alexiadou, N., Garaz, S., Gonzalez-Monteagudo, J., & Taba, M. (2018). Internationalisation and migrant academics: The hidden narratives of mobility. Higher Education, 76(3), 537-554. https://doi.org/10.1007/s10734-017-0224-z Mormina, M., & Istratii, R. (2021). 'Capacity for what? Capacity for whom?' A decolonial deconstruction of research capacity development practices in the Global South and a proposal for a value-centred approach. Wellcome Open Research, 6, Article 129. https://doi.org/10.12688/wellcomeopenres.16850.1 Morrison, T. H., Adger, W. N., Brown, K., Lemos, M. C., Huitema, D., Phelps, J., Evans, L., Cohen, P., Song, A. M., Turner, R., Quinn, T., & Hughes, T. P. (2019). The black box of power in polycentric environmental governance. Global Environmental Change, 57, 101934. https://doi.org/10.1016/j.gloenvcha.2019.101934 Morton, B., Vercueil, A., Masekela, R., Heinz, E., Reimer, L., Saleh, S., Kalinga, C., Seekles, M., Biccard, B., Chakaya, J., Abimbola, S., Obasi, A., & Oriyo, N. (2022). Consensus statement on measures to promote equitable authorship in the publication of research from international partnerships. Anaesthesia, 77(3), 264-276. https://doi.org/10.1111/anae.15597 Morvaridi, B., & Hughes, C. (2018). South-South cooperation and neoliberal hegemony in a post-aid world. Development and Change, 49(3), 867-892. https://doi.org/10.1111/dech.12405 Moshtari, M., & Safarpour, A. (2024). Challenges and strategies for the internationalization of higher education in low-income East African countries. Higher Education, 87(1), 89-109. https://doi.org/10.1007/s10734-023-00994-1 Ndlovu-Gatsheni, S. J. (2021). Internationalisation of higher education for pluriversity: A decolonial reflection. Journal of the British Academy, 9(s1), 77-98. https://doi.org/10.5871/jba/009s1.077 Nkansah, J. O., Oldac, Y. I., & Yang, L. (2025). Dependency and neocolonialism in international research collaboration: Evidence from a Ghanaian elite university. Higher Education, 90(6), 1571-1589. https://doi.org/10.1007/s10734-024-01393-w Nonet, G. A.-H., Gössling, T., Van Tulder, R., & Bryson, J. M. (2022). Multi-stakeholder engagement for the Sustainable Development Goals: Introduction to the special issue. Journal of Business Ethics, 180(4), 945-957. https://doi.org/10.1007/s10551-022-05192-0 Pashby, K., & Andreotti, V. (2016). Ethical internationalisation in higher education: Interfaces with international development and sustainability. Environmental Education Research, 22(6), 771-787. https://doi.org/10.1080/13504622.2016.1201789 Pattberg, P., & Widerberg, O. (2016). Transnational multistakeholder partnerships for sustainable development: Conditions for success. Ambio, 45(1), 42-51. https://doi.org/10.1007/s13280-015-0684-2 Purnell, P. J. (2024). Geodiversity of research: Geographical topic focus, author location, and collaboration. A case study of SDG 2: Zero hunger. Scientometrics, 129(5), 2701-2727. https://doi.org/10.1007/s11192-024-04994-5 Ramírez-Castañeda, V. (2020). Disadvantages in preparing and publishing scientific papers caused by the dominance of the English language in science: The case of Colombian researchers in biological sciences. PLoS ONE, 15(9), e0238372. https://doi.org/10.1371/journal.pone.0238372 Ran, B., & Qi, H. (2018). Contingencies of power sharing in collaborative governance. The American Review of Public Administration, 48(8), 836-851. https://doi.org/10.1177/0275074017745355 Robertson, S. L., & Komljenović, J. (2016). Non-state actors, and the advance of frontier higher education markets in the global south. Oxford Review of Education, 42(5), 594-611. https://doi.org/10.1080/03054985.2016.1224302 Schroeder, D., Chatfield, K., Singh, M., Chennells, R., & Herissone-Kelly, P. (2019). Equitable research partnerships: A global code of conduct to counter ethics dumping. Springer. https://doi.org/10.1007/978-3-030-15745-6 Skrivankova, V. W., Hossmann, S., Cornell, M., Ballif, M., Dupont, C., Huwa, J., Seintaridis, K., Kalua, T., Wandeler, G., Kassanjee, R., Haas, A. D., Technau, K.-G., Fenner, L., Low, N., Davies, M.-A., & Egger, M. (2023). Authorship inequalities in global health research: The IeDEA Southern Africa collaboration. BMJ Global Health, 8(12), e013316. https://doi.org/10.1136/bmjgh-2023-013316 Skute, I., Zalewska-Kurek, K., Hatak, I., & de Weerd-Nederhof, P. (2019). Mapping the field: A bibliometric analysis of the literature on university-industry collaborations. The Journal of Technology Transfer, 44(3), 916-947. https://doi.org/10.1007/s10961-017-9637-1 Staffa, R. K., Riechers, M., & Martín-López, B. (2022). A feminist ethos for caring knowledge production in transdisciplinary sustainability science. Sustainability Science, 17(1), 45-63. https://doi.org/10.1007/s11625-021-01064-0 Stein, S. (2016). Rethinking the ethics of internationalization: Five challenges for higher education. InterActions: UCLA Journal of Education and Information Studies, 12(2). https://doi.org/10.5070/d4122031205 Stein, S., Andreotti, V., Bruce, J., & Susa, R. (2016). Towards different conversations about the internationalization of higher education. Comparative and International Education, 45(1). https://doi.org/10.5206/cie-eci.v45i1.9281 Tight, M. (2021). Globalization and internationalization as frameworks for higher education research. Research Papers in Education, 36(1), 52-74. https://doi.org/10.1080/02671522.2019.1633560 Tikly, L. (2019). Education for sustainable development in Africa: A critique of regional agendas. Asia Pacific Education Review, 20(2), 223-237. https://doi.org/10.1007/s12564-019-09600-5 Turner, R. A., Addison, J., Arias, A., Bergseth, B. J., Marshall, N. A., Morrison, T. H., & Tobin, R. C. (2016). Trust, confidence, and equity affect the legitimacy of natural resource governance. Ecology and Society, 21(3), Article 18. https://doi.org/10.5751/es-08542-210318 Van der Veken, K., Belaid, L., Delvaux, T., & De Brouwere, V. (2017). Research capacity building through North-South-South networking: Towards true partnership? An exploratory study of a network for scientific support in the field of sexual and reproductive health. Health Research Policy and Systems, 15, Article 39. https://doi.org/10.1186/s12961-017-0202-z West, S., van Kerkhoff, L., & Wagenaar, H. (2019). Beyond 'linking knowledge and action': Towards a practice-based approach to transdisciplinary sustainability interventions. Policy Studies, 40(5), 534-555. https://doi.org/10.1080/01442872.2019.1618810 Wondimagegn, D., Whitehead, C., Cartmill, C., Rodrigues, E., Correia, A., Lins, T. S., & Costa, M. J. (2023). Faster, higher, stronger - Together? A bibliometric analysis of author distribution in top medical education journals. BMJ Global Health, 8(6), e011656. https://doi.org/10.1136/bmjgh-2022-011656 Hashtags: #PartnershipGovernance #InternationalisationOfHigherEducation #CapacityBuilding #EpistemicJustice #ResearchCollaboration #Accountability #GlobalPartnerships #MultiStakeholder #KnowledgeExchange #CrossSectoral #ParticipatorySettlement #AllocativeInstruments #ProceduralInstruments #EpistemicAuthority #ResourceCustody #SDG17

Author: Fatima Al Mansouri Affiliation: Swiss International University (SIU) ORCID ID: 0009-0003-3004-395X Submitted 10 April 2026; Revised 05 June 2026; Accepted 20 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y10028 Volume 3, December 2026, (10028) Abstract Indoor plants are widely seen as harmless additions to homes, offices and schools, valued for oxygen, air purification and comfort. This article argues that the health question is more complex. In ordinary rooms, oxygen depletion by potted plants is unlikely to be a meaningful risk, because plant respiration is small relative to room air volume and human respiration. More relevant risks arise from conditions plants can create or intensify: wet soil, poor drainage, damp surfaces, microbial growth, dust and allergens. Combining a literature review with an 18-month comparison of ten educational institutions (five with indoor plants, five with greenery outdoors only), the study examined breathing comfort, moisture, maintenance and absence. The findings do not support a causal claim that plants alone determine health or performance, but plant-free indoor spaces were easier to clean and keep dry, and more consistently linked to comfortable breathing. The article shifts attention from the oxygen myth to the practical governance of moisture, ventilation and maintenance, drawing on symbolic capital and institutional isomorphism to explain why schools adopt indoor planting despite uncertain health value. It concludes that schools should prioritize ventilation, dry surfaces, humidity control and outdoor greenery over permanent indoor planting. Keywords: Indoor plants; Indoor air quality; School health; Respiratory risk; Dampness and mold; Ventilation 1. Introduction Indoor plants occupy a familiar place in homes, offices, schools, clinics, hotels and public institutions. They are visually associated with freshness, care, environmental awareness and comfort. In schools, plants may also have educational value, because they can support lessons about growth, ecology and responsibility. These design and pedagogical functions are legitimate, but they do not settle the environmental-health question. A classroom is not only a decorated room; it is a shared breathing environment used for long periods by children, teachers and staff, including people with asthma, rhinitis, allergies or other respiratory sensitivities. The most common justification for indoor plants is that they add oxygen and purify indoor air. This argument is built on a biological truth: in sufficient light, plants photosynthesize and release oxygen. The argument becomes weaker when it is transferred to ordinary rooms, where light may be poor, ventilation may be variable and plants also respire. Respiration occurs continuously, including at night and in low-light conditions. The amount of oxygen consumed by typical potted plants is nonetheless very small in relation to the oxygen available in a normal room. A health argument focused mainly on oxygen depletion therefore risks overstating a minor mechanism while ignoring more plausible risks. The more important concern is the microenvironment around the plant. Potted plants usually require soil, watering and organic matter. When soil remains wet, pots drain poorly, dead leaves accumulate or plants are placed on porous surfaces, they may contribute to dampness, microbial activity, dust and odors. These factors are particularly relevant in schools, where rooms may be crowded, cleaning routines may be stretched, windows may remain closed for climate or pollution reasons, and children may touch soil or leaves. Indoor environmental quality in schools has been linked to health symptoms, absence and learning conditions in a substantial body of research (Daisey et al., 2003; Sadrizadeh et al., 2022; Wargocki et al., 2020; Wargocki and Wyon, 2017). Dampness and mold exposure have also been repeatedly associated with respiratory and allergic outcomes (Caillaud et al., 2018; Fisk et al., 2007; Fisk et al., 2019; Jaakkola et al., 2013; Mendell et al., 2011; Mudarri and Fisk, 2007; Quansah et al., 2012). The research gap addressed here is not whether plants can look pleasant or whether vegetation has value. The gap is narrower and more practical. The literature often treats plant benefits, dampness risks, ventilation research and institutional symbolism separately, but it rarely asks whether permanent decorative planting inside educational buildings is a prudent indoor-air practice once maintenance capacity, respiratory sensitivity and shared occupancy are taken seriously. Existing plant-air-quality studies also caution that chamber findings cannot be generalized directly to real buildings (Cummings and Waring, 2020). This article therefore examines indoor plants as part of a wider institutional and environmental system rather than as isolated decorative objects. The article has three aims. First, it distinguishes the oxygen myth from more credible moisture and respiratory risks. Second, it reports an 18-month observational comparison of educational institutions with and without permanent indoor plants. Third, it develops a theoretical explanation for why schools may adopt indoor plants as a symbol of care and sustainability even when the health case is uncertain. 2. Background and Theoretical Framework 2.1 Plants and oxygen balance Plant physiology matters, but it should be interpreted proportionately. Photosynthesis depends on adequate light, while respiration occurs continuously. A medium-sized room may contain thousands of liters of oxygen; the amount used by one or several potted plants overnight is small compared with this volume and much smaller than the oxygen demand of the human occupants. This does not mean that plants function as meaningful oxygen-management systems. It means that both popular claims should be treated carefully: ordinary indoor plants are unlikely to create direct oxygen deficiency, but they also should not be presented as substitutes for ventilation. Indoor air quality is governed mainly by source control, ventilation, filtration, humidity control and maintenance. Research on schools shows that classroom ventilation is frequently inadequate and that better ventilation is associated with reduced illness absence and stronger learning conditions (Bako-Biro et al., 2012; Fisk, 2017; Haverinen-Shaughnessy et al., 2011; Haverinen-Shaughnessy and Shaughnessy, 2015; Mendell et al., 2013). Studies of office environments also show that ventilation, carbon dioxide and volatile organic compounds can affect cognitive performance and perceived indoor quality (Allen et al., 2016; Satish et al., 2012). Against this evidence, decorative plants should be understood as a minor and highly context-dependent element, not as a primary air-quality intervention. 2.2 Moisture, mold and bioaerosols The strongest health concern is not oxygen but damp organic material. Indoor dampness and mold are consistently associated with respiratory and allergic symptoms across multiple reviews and meta-analyses (Caillaud et al., 2018; Fisk et al., 2007; Fisk et al., 2019; Jaakkola et al., 2013; Kanchongkittiphon et al., 2015; Mendell et al., 2011; Mudarri and Fisk, 2007; Quansah et al., 2012). Schools and day-care settings are especially important, because children and staff spend long hours in shared rooms and because maintenance problems can persist unnoticed. Bioaerosols are dynamic: biological particles can be emitted, resuspended, deposited and removed through ventilation, cleaning and filtration (Cox et al., 2020; Nazaroff, 2016). Wet soil, decaying leaves and dusty surfaces may not automatically create harmful exposure, but they add unnecessary complexity to environments that should be easy to clean and keep dry. 2.3 Indoor plants and air-cleaning claims The idea that indoor plants purify air became prominent partly because sealed-chamber studies showed that plants, or plant-associated microorganisms, could remove selected volatile organic compounds. Real buildings are different. Doors open, people move, ventilation dilutes pollutants, cleaning products emit chemicals and pollutant sources vary over time. Reviews have therefore questioned whether potted plants can meaningfully improve indoor air quality under ordinary building conditions (Cummings and Waring, 2020; Han and Ruan, 2020; Ravindra et al., 2022). Recent evidence suggests that plants may influence relative humidity more than carbon dioxide or temperature in some office settings (Jiang et al., 2024). In a dry climate this may sometimes be perceived as comfortable, but in poorly ventilated or moisture-sensitive rooms, added humidity can be undesirable. 2.4 Symbolic capital Bourdieu’s concept of symbolic power helps explain why indoor plants retain institutional appeal (Bourdieu, 1989). A green reception area or a plant-filled classroom can signal care, modernity, environmental awareness and taste. These symbolic meanings are socially powerful, because visitors, parents and staff may interpret greenery as evidence of wellbeing. The problem is not the symbolism itself; it arises when symbolic health is mistaken for measured or well-managed health. 2.5 Institutional isomorphism DiMaggio and Powell’s theory of institutional isomorphism explains why practices spread even when their technical value is uncertain (DiMaggio and Powell, 1983). Schools may copy the green corners, biophilic design language or reception-area plants used by other institutions. Over time, a decorative practice can become normalized, and its maintenance burden may be overlooked. This article therefore treats indoor planting as both an environmental object and an institutional practice. 2.6 Equity and context Indoor plant risk is not uniform. A bright, spacious, well-ventilated building with trained facility staff is different from a crowded, humid, under-maintained one. Hot climates may keep windows closed because of air conditioning; cold climates may keep them closed for heat retention; polluted urban environments may keep them closed to avoid outdoor pollutants. In each case, the same potted plant may carry different implications for moisture, cleaning and user comfort. A precautionary school policy should therefore consider local climate, ventilation, occupant vulnerability and maintenance capacity. 3. Methodology 3.1 Research design The study used a qualitative comparative case-study design, supported by an observational component and a focused review of the environmental-health literature. The design was appropriate because the research question concerns institutional practice, maintenance conditions and perceived indoor comfort rather than clinical diagnosis. The study did not seek to isolate a single biological cause; it examined whether permanent indoor plants formed part of a broader pattern of indoor environmental risk in educational settings. 3.2 Case selection Ten schools or educational institutes were observed over 18 months. The cases were selected to create a practical contrast between two institutional approaches. Group A included five institutions that kept potted plants inside classrooms, corridors, reception areas, staff rooms or shared learning spaces. Group B included five institutions that maintained plants outside only, such as in gardens, entrance landscaping, courtyards or exterior planters, without permanent soil-based plants in indoor learning spaces. The comparison was designed to preserve the original observational logic while improving analytical clarity: the unit of analysis was the institution as an indoor environmental system, not the individual plant. Table 1 summarizes the characteristics of the two groups. Table 1. Characteristics of the two institutional groups. Feature Group A – Indoor-plant institutions Group B – Plant-free indoor institutions Number of institutions 5 5 Plant location Inside classrooms, corridors, reception areas, staff rooms and shared learning spaces Outdoors only: gardens, entrance landscaping, courtyards and exterior planters Permanent indoor soil-based plants Present Absent Observation period 18 months 18 months Unit of analysis Institution as an indoor environmental system Institution as an indoor environmental system 3.3 Scope and data sources Data consisted of repeated site observations, field notes on visible moisture and maintenance conditions, informal non-identifiable feedback from staff where available, and institutional-level attendance and learning-stability impressions where institutions could share such information in general terms. No medical examinations, lung-function tests, personal health records or laboratory air samples were collected. The analysis therefore does not make clinical claims. It reports institutional patterns and interprets them in light of established evidence on dampness, mold, ventilation and classroom air quality. 3.4 Observation focus The observations focused on six domains, set out in Table 2: the location and density of indoor plants; watering and drainage conditions; visible signs of dampness, staining, dead leaves, odor, insects or dust; the cleanability of surfaces around pots; perceived breathing comfort in classrooms and corridors; and general patterns of absence or learning continuity, as described or observed at the institutional level. Attention was also given to plausible confounders, including building age, crowding, ventilation habits, cleaning routines, climate conditions, management discipline and room use. Teacher-reported building-related symptoms and child respiratory symptoms in moisture-damaged schools provide relevant support for including these domains in school observations (Casas et al., 2017; Kielb et al., 2015). Table 2. Observation domains and the indicators recorded for each domain. Observation domain Indicators recorded 1. Location and density of indoor plants Placement (classrooms, corridors, reception, staff rooms); number and concentration of pots per space 2. Watering and drainage Watering regularity; standing water; drainage adequacy; presence of an assigned caretaker 3. Visible dampness and contamination Damp soil, water marks, staining, dead leaves, musty odor, insects, dust accumulation 4. Cleanability of surfaces Porous vs. non-porous surfaces around pots; clutter; ease of inspection and cleaning 5. Perceived breathing comfort Staff and student descriptions of air as clear, dry, heavy or musty near plant locations 6. Absence and learning continuity Institution-level patterns of attendance and learning stability described or observed over time Confounders considered Building age, crowding, ventilation habits, cleaning routines, climate, management discipline, room use 3.5 Analytical procedure The analysis used cross-case pattern matching. First, the field notes from the indoor-plant group were compared with those from the plant-free indoor group. Second, repeated observations over the 18-month period were distinguished from one-time events. Third, the emerging patterns were interpreted conservatively against the literature. A pattern was treated as meaningful only when it appeared repeatedly or coherently across cases; it was not treated as proof of causation. This conservative approach is important because absence and academic performance are shaped by many factors beyond the indoor environment. 3.6 Ethical scope The study relied on non-identifiable institutional observations and did not collect personal medical data, individual student records or identifiable participant responses. It should therefore be read as an environmental and organizational case comparison rather than as human-subject clinical research. 4. Analysis and Findings This section reports the cross-case patterns observed over the 18-month period. Table 3 summarizes the comparative observations across the six domains for the two groups, and Figure 1 presents these patterns as a graphical summary. The detailed findings and their associated propositions follow, and are consolidated in Table 4. Table 3. Comparative summary of observations across the two groups over the 18-month period. Observation domain Group A – Indoor-plant institutions Group B – Plant-free indoor institutions Oxygen-related risk No indication of measurable oxygen depletion; oxygen claims sometimes used to justify planting No indication of measurable oxygen depletion; ventilation framed as the main air concern Moisture and maintenance Recurrent damp soil, inconsistent watering, water marks, dead leaves and occasional musty odors Fewer soil-based and water-retaining objects; simpler moisture control Cleanability of spaces More cluttered corners, organic residues and objects requiring specialized maintenance Fewer cluttered corners and organic residues; easier inspection and cleaning Perceived breathing comfort Occasional reports of heaviness, musty smell or discomfort near plant locations Rooms more often described as clear, dry and easier to breathe in when ventilated Absence and learning continuity No causal pattern identified; influenced by many external factors Cautiously more stable attendance and learning routines during parts of the period Symbolic value Plants prominent in visible spaces, signaling care; able to mask maintenance problems Care signaled through outdoor greenery and tidy, dry indoor spaces Figure 1. Graphical summary of the comparison between indoor-plant and plant-free indoor learning spaces, highlighting the shift in emphasis from the oxygen myth toward moisture, maintenance and respiratory comfort. 4.1 Oxygen was not the main risk The first finding is that the oxygen argument is often misunderstood. Plants do respire and therefore use oxygen, particularly when photosynthesis is limited. In normal indoor settings, however, oxygen use by ordinary potted plants is too small to explain meaningful oxygen deficiency. The more important implication is educational and managerial: the belief that plants oxygenate rooms can create false confidence and distract schools from ventilation, source control and moisture management (Cummings and Waring, 2020; Fisk, 2017). Proposition 1. In ordinary educational rooms, indoor plants are unlikely to create direct oxygen depletion, but oxygen-related claims can mislead institutions when they are used to justify permanent planting without adequate ventilation and maintenance. 4.2 Moisture and maintenance were the strongest practical concerns The indoor-plant institutions showed more visible plant-related maintenance issues. These included damp soil, inconsistent watering, water marks near pots, dead leaves left in containers, plants placed near porous materials and occasional musty odors. These observations do not prove harmful exposure, but they are consistent with the literature identifying dampness and mold as preventable respiratory risk factors (Caillaud et al., 2018; Fisk et al., 2007; Fisk et al., 2019; Jaakkola et al., 2013; Kanchongkittiphon et al., 2015; Mendell et al., 2011). The plant-free indoor institutions had fewer soil-based and water-retaining objects in learning areas, which made cleaning and moisture control simpler. Proposition 2. Where maintenance is inconsistent, permanent indoor plants can become avoidable moisture-management liabilities in schools, especially when soil, pots and surrounding surfaces remain damp. 4.3 Plant-free indoor spaces were easier to keep clean and dry The comparison suggested that classrooms and corridors without potted plants had fewer cluttered corners, fewer organic residues and fewer objects requiring specialized maintenance. Cleanability is a practical indoor-health variable. It is not only a matter of visual tidiness; it affects dust accumulation, inspection routines and the ability to identify dampness quickly. This finding aligns with the broader principle that source control and maintenance are central to indoor environmental quality (Cox et al., 2020; Daisey et al., 2003; Nazaroff, 2016; Sadrizadeh et al., 2022). Proposition 3. Plant-free indoor learning spaces may support better environmental control because they reduce unnecessary damp organic material and simplify cleaning, inspection and maintenance routines. 4.4 Perceived breathing comfort favored the plant-free indoor group Across the observation period, staff and students in plant-free indoor settings more often described rooms as clear, dry or easier to breathe in, particularly when the rooms were ventilated and uncluttered. In the indoor-plant group, feedback occasionally mentioned heaviness, a musty smell or discomfort near plant locations. These reports are subjective and cannot be treated as clinical outcomes. They are nevertheless relevant, because perceived air quality influences comfort and may shape attention, fatigue and satisfaction in educational settings (Allen et al., 2016; Satish et al., 2012; Wargocki et al., 2020; Wargocki and Wyon, 2017). Proposition 4. Plant-free indoor policies may improve perceived breathing comfort when combined with ventilation, dry surfaces and disciplined cleaning routines. 4.5 Absence and learning stability showed a cautious favorable pattern The plant-free indoor institutions appeared to show more stable attendance and learning routines during parts of the observation period. This finding must be stated carefully. Absence is affected by infections, family decisions, transport, seasonality, school policy, socioeconomic background and many other variables. Academic performance is affected by teaching, curriculum, leadership and student background. The study therefore does not claim that removing plants directly reduces absence or raises grades. The more defensible interpretation is that simpler, cleaner and drier indoor environments may support attendance and learning conditions indirectly, which is consistent with research linking classroom ventilation and air quality with absence and performance (Bako-Biro et al., 2012; Haverinen-Shaughnessy et al., 2011; Haverinen-Shaughnessy and Shaughnessy, 2015; Mendell et al., 2013; Wargocki et al., 2020). Proposition 5. Plant-free indoor environments should be understood as one component of a broader preventive strategy that may support attendance and learning continuity, rather than as an independent determinant of academic outcomes. 4.6 Indoor plants carried symbolic value The indoor-plant institutions often used plants in visible spaces such as entrances, corridors and shared rooms. In these settings, plants communicated care, naturalness and institutional warmth. This symbolic value is understandable and may improve first impressions. However, it can also obscure maintenance risk. A plant can look healthy while its soil is damp, its leaves are dusty or its pot is leaking. The symbolic analysis matters because many school decisions are not purely technical; they are shaped by what appears modern, caring and legitimate (Bourdieu, 1989; Bringslimark et al., 2009; DiMaggio and Powell, 1983). Proposition 6. Indoor plants can function as symbolic capital in educational institutions, but symbolic indicators of care should not be allowed to replace evidence-based indoor environmental management. Table 4. Summary of the main findings and their associated analytical propositions. Finding Key observation Proposition 4.1 Oxygen No measurable oxygen depletion; oxygen claims can create false confidence P1: Oxygen claims can mislead when used to justify planting without ventilation 4.2 Moisture Recurrent damp soil, water marks, dead leaves and musty odors in the plant group P2: Indoor plants can become avoidable moisture liabilities where maintenance is weak 4.3 Cleanability Plant-free spaces had fewer residues and were easier to inspect and clean P3: Plant-free spaces simplify cleaning, inspection and moisture control 4.4 Comfort Plant-free rooms more often described as clear and easier to breathe in P4: Plant-free policies may improve perceived comfort alongside ventilation 4.5 Absence/learning Cautiously more stable attendance in plant-free settings P5: Plant-free spaces are one component of a broader preventive strategy 4.6 Symbolism Plants signaled care but could mask maintenance problems P6: Symbolic care should not replace evidence-based management 5. Discussion The findings support a cautious and proportionate position. Indoor plants should not be described as inherently dangerous. A small, well-maintained plant in a bright, ventilated, non-crowded area may create little risk. The stronger argument is not prohibition in every possible setting but prevention in high-occupancy learning spaces, where children or sensitive users share air for long periods. In such spaces, avoidable sources of damp soil, dust and microbial growth should be minimized unless there is a clear educational purpose and a clear maintenance protocol. The article contributes to indoor environmental quality research by reframing the indoor-plant debate. Much public discussion treats plants either as oxygen producers or as natural air purifiers. This framing is too narrow. The more useful framework is environmental governance: who waters the plant, how drainage is managed, whether surfaces can be cleaned, whether soil remains wet, whether leaves collect dust, whether humidity is monitored, and whether the room has adequate air exchange. This shift moves the debate from plant symbolism to facility-management responsibility. The article also contributes to school-health debates by connecting respiratory risk with everyday design practices. Research on ventilation, dampness and classroom air quality already shows that learning environments are health environments (Daisey et al., 2003; Fisk, 2017; Sadrizadeh et al., 2022; Wargocki et al., 2020). The present study adds a practical micro-level issue: decorative soil-based objects can become part of the indoor environmental burden when they are unmanaged. This contribution is modest but important, because many school risks are not dramatic hazards; they are small, repeated, preventable conditions that accumulate through routine neglect. The theoretical contribution is to show how symbolic capital and institutional isomorphism help explain why potentially weak practices persist. Plants may be adopted because they make schools look caring, ecological and modern. Once adopted by many institutions, the practice becomes normal and may escape scrutiny. This does not mean that all symbolic practices are harmful. It means that institutional symbols should be tested against health, maintenance and equity criteria. A healthy school is not defined by how green it looks indoors, but by whether its indoor spaces are dry, clean, ventilated and safe for vulnerable users. The findings also refine debates on biophilic design. Exposure to nature and greenery can support wellbeing, and the psychological literature recognizes potential benefits of indoor plants in some contexts (Bringslimark et al., 2009). The present argument does not reject greenery; it relocates greenery. Outdoor gardens, courtyards, balcony planting, supervised short-term classroom experiments and exterior landscaping can support ecological education while reducing permanent indoor moisture sources. This distinction matters, because it allows schools to preserve environmental learning without treating permanent indoor planting as a health intervention. 6. Practical Implications • Schools and institutes should avoid using indoor plants as an air-cleaning or oxygen-management strategy. Ventilation, filtration, humidity control, low-emission materials and systematic cleaning are more reliable measures. • Permanent indoor plants should be avoided in classrooms, nurseries, sleeping rooms, libraries with poor ventilation, crowded corridors, carpeted areas and rooms used by children with known respiratory sensitivity. • Where plants are kept indoors for limited decorative purposes, they should be few in number, placed on non-porous cleanable surfaces, kept away from carpets and books, watered only under assigned responsibility and removed immediately if there is mold, odor, insects, water leakage or decaying material. • Environmental education should prioritize outdoor gardens, exterior planting, supervised laboratory activities and temporary experiments with clear cleanup procedures. • Schools should treat humidity, dampness and ventilation as routine governance issues, not as problems to be addressed only after visible mold or complaints appear. 7. Limitations and Future Research This study has limitations. It was observational and qualitative, not randomized or clinical. The number of institutions was small, and the cases were not statistically representative. The study did not include air sampling, microbial identification, lung-function testing, medical records or standardized psychometric measures of perceived air quality. Absence and academic performance were interpreted cautiously, because they are influenced by many institutional, social and seasonal factors. The findings should therefore be understood as analytical and preventive propositions rather than as causal estimates. Future research should test these propositions with larger mixed-method designs. Useful studies would combine standardized building inspections, humidity monitoring, ventilation-rate measurement, microbial sampling, occupant symptom surveys and anonymized attendance data. Experimental or quasi-experimental studies could examine schools before and after removing permanent indoor plants while controlling for cleaning, ventilation and season. Further research should also distinguish between climates, building types, plant species, soil media, watering protocols and room uses. Such work would help move the debate from general claims about plants to evidence-based guidance for specific educational environments. 8. Conclusions Indoor plants are not a major direct oxygen-depletion risk in ordinary rooms, but the oxygen debate is the wrong center of attention. In schools and other shared learning spaces, the more credible concern is moisture, mold, microbial activity, dust, allergens and uneven maintenance. The 18-month comparison reported here suggests that plant-free indoor learning spaces can be easier to keep clean, dry and comfortable, while outdoor greenery can still support environmental learning. The contribution of this article is both practical and theoretical. Practically, it recommends a precautionary shift from decorative indoor planting toward ventilation, humidity control, dry surfaces and outdoor greenery. Theoretically, it shows how indoor plants can operate as symbols of care and modernity even when their health value is uncertain. For educational institutions, the safest message is simple: greenery is valuable, but shared indoor air should be governed by evidence, maintenance capacity and respiratory protection rather than by appearance alone. Funding No specific funding was received for this research from public, commercial or non-profit organizations. Data Availability Statement The study is based on non-identifiable institutional observations and literature analysis. No personal medical data or identifiable participant data were collected. Additional raw field notes are not publicly shared, because they could indirectly identify the institutions. Ethics Statement The study used non-identifiable institutional observations and did not involve clinical testing, intervention, personal health records or identifiable human-subject data. Formal ethics approval was therefore not required under the scope of this environmental and organizational case analysis. Declaration of Competing Interest The author declare no conflict of interest. Declaration on the Use of Artificial Intelligence Artificial intelligence tools were used only to improve the language, style and editing of the manuscript. The ideas, theory, analysis, interpretation and final decisions were made by the author. The author take full responsibility for the content and integrity of the manuscript. References Allen JG, MacNaughton P, Satish U, Santanam S, Vallarino J, Spengler JD. Associations of cognitive function scores with carbon dioxide, ventilation, and volatile organic compound exposures in office workers: A controlled exposure study of green and conventional office environments. Environmental Health Perspectives 2016;124(6):805-12. https://doi.org/10.1289/ehp.1510037 Bako-Biro Z, Clements-Croome DJ, Kochhar N, Awbi HB, Williams MJ. Ventilation rates in schools and pupils’ performance. Building and Environment 2012;48:215-23. https://doi.org/10.1016/j.buildenv.2011.08.018 Bourdieu P. Social space and symbolic power. Sociological Theory 1989;7(1):14-25. https://doi.org/10.2307/202060 Bringslimark T, Hartig T, Patil GG. The psychological benefits of indoor plants: A critical review of the experimental literature. Journal of Environmental Psychology 2009;29(4):422-33. https://doi.org/10.1016/j.jenvp.2009.05.001 Caillaud D, Leynaert B, Keirsbulck M, Nadif R. Indoor mould exposure, asthma and rhinitis: Findings from systematic reviews and recent longitudinal studies. European Respiratory Review 2018;27(148):Article No. 170137. https://doi.org/10.1183/16000617.0137-2017 Casas L, Tischer C, Wouters IM, Valkonen M, Gehring U, Doekes G, et al. School attendance and daily respiratory symptoms in children: Influence of moisture damage. Indoor Air 2017;27(2):303-10. https://doi.org/10.1111/ina.12311 Cox J, Mbareche H, Lindsley WG, Duchaine C. Field sampling of indoor bioaerosols. Aerosol Science and Technology 2020;54(5):572-84. https://doi.org/10.1080/02786826.2019.1688759 Cummings BE, Waring MS. Potted plants do not improve indoor air quality: A review and analysis of reported VOC removal efficiencies. Journal of Exposure Science and Environmental Epidemiology 2020;30:253-61. https://doi.org/10.1038/s41370-019-0175-9 Daisey JM, Angell WJ, Apte MG. Indoor air quality, ventilation and health symptoms in schools: An analysis of existing information. Indoor Air 2003;13(1):53-64. https://doi.org/10.1034/j.1600-0668.2003.00153.x DiMaggio PJ, Powell WW. The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review 1983;48(2):147-60. https://doi.org/10.2307/2095101 Fisk WJ. The ventilation problem in schools: Literature review. Indoor Air 2017;27(6):1039-51. https://doi.org/10.1111/ina.12403 Fisk WJ, Chan WR, Johnson AL. Does dampness and mold in schools affect health? Results of a meta-analysis. Indoor Air 2019;29(6):895-902. https://doi.org/10.1111/ina.12588 Fisk WJ, Lei-Gomez Q, Mendell MJ. Meta-analyses of the associations of respiratory health effects with dampness and mold in homes. Indoor Air 2007;17(4):284-96. https://doi.org/10.1111/j.1600-0668.2007.00475.x Han KT, Ruan LW. Effects of indoor plants on air quality: A systematic review. Environmental Science and Pollution Research 2020;27:16019-51. https://doi.org/10.1007/s11356-020-08174-9 Haverinen-Shaughnessy U, Moschandreas DJ, Shaughnessy RJ. Association between substandard classroom ventilation rates and students’ academic achievement. Indoor Air 2011;21(2):121-31. https://doi.org/10.1111/j.1600-0668.2010.00686.x Haverinen-Shaughnessy U, Shaughnessy RJ. Effects of classroom ventilation rate and temperature on students’ test scores. PLoS ONE 2015;10(8):Article No. e0136165. https://doi.org/10.1371/journal.pone.0136165 Jaakkola MS, Quansah R, Hugg TT, Heikkinen SAM, Jaakkola JJK. Association of indoor dampness and molds with rhinitis risk: A systematic review and meta-analysis. Journal of Allergy and Clinical Immunology 2013;132(5):1099-110. https://doi.org/10.1016/j.jaci.2013.07.028 Jiang J, Irga P, Coe R, Gibbons P. Effects of indoor plants on CO2 concentration, indoor air temperature and relative humidity in office buildings. PLoS ONE 2024;19(7):Article No. e0305956. https://doi.org/10.1371/journal.pone.0305956 Kanchongkittiphon W, Mendell MJ, Gaffin JM, Wang G, Phipatanakul W. Indoor environmental exposures and exacerbation of asthma: An update to the 2000 review by the Institute of Medicine. Environmental Health Perspectives 2015;123(1):6-20. https://doi.org/10.1289/ehp.1307922 Kielb C, Lin S, Muscatiello N, Hord W, Rogers-Harrington J, Healy J. Building-related health symptoms and classroom indoor air quality: A survey of school teachers in New York State. Indoor Air 2015;25(4):371-80. https://doi.org/10.1111/ina.12154 Mendell MJ, Eliseeva EA, Davies MM, Spears M, Lobscheid A, Fisk WJ, et al. Association of classroom ventilation with reduced illness absence: A prospective study in California elementary schools. Indoor Air 2013;23(6):515-28. https://doi.org/10.1111/ina.12042 Mendell MJ, Mirer AG, Cheung K, Tong M, Douwes J. Respiratory and allergic health effects of dampness, mold, and dampness-related agents: A review of the epidemiologic evidence. Environmental Health Perspectives 2011;119(6):748-56. https://doi.org/10.1289/ehp.1002410 Mudarri D, Fisk WJ. Public health and economic impact of dampness and mold. Indoor Air 2007;17(3):226-35. https://doi.org/10.1111/j.1600-0668.2007.00474.x Nazaroff WW. Indoor bioaerosol dynamics. Indoor Air 2016;26(1):61-78. https://doi.org/10.1111/ina.12174 Quansah R, Jaakkola MS, Hugg TT, Heikkinen SAM, Jaakkola JJK. Residential dampness and molds and the risk of developing asthma: A systematic review and meta-analysis. PLoS ONE 2012;7(11):Article No. e47526. https://doi.org/10.1371/journal.pone.0047526 Ravindra K, Kaur-Sidhu M, Mor S, John S. Phytoremediation potential of indoor plants in reducing air pollutants. Frontiers in Sustainable Cities 2022;4:Article No. 1039710. https://doi.org/10.3389/frsc.2022.1039710 Sadrizadeh S, Yao R, Yuan F, Awbi H, Bahnfleth W, Bi Y, et al. Indoor air quality and health in schools: A critical review for developing the roadmap for the future school environment. Journal of Building Engineering 2022;57:Article No. 104908. https://doi.org/10.1016/j.jobe.2022.104908 Satish U, Mendell MJ, Shekhar K, Hotchi T, Sullivan D, Streufert S, et al. Is CO2 an indoor pollutant? Direct effects of low-to-moderate CO2 concentrations on human decision-making performance. Environmental Health Perspectives 2012;120(12):1671-7. https://doi.org/10.1289/ehp.1104789 Wargocki P, Porras-Salazar JA, Contreras-Espinoza S, Bahnfleth W. The relationships between classroom air quality and children’s performance in school. Building and Environment 2020;173:Article No. 106749. https://doi.org/10.1016/j.buildenv.2020.106749 Wargocki P, Wyon DP. Ten questions concerning thermal and indoor air quality effects on the performance of office work and schoolwork. Building and Environment 2017;112:359-66. https://doi.org/10.1016/j.buildenv.2016.11.020 #IndoorPlants #IndoorAirQuality #SchoolHealth #RespiratoryRisk #DampnessAndMold #Ventilation #EnvironmentalHealth #PublicHealth #U7YJournal #U7Y #AcademicResearch #OpenAccess

Author: Sofia García Affiliation: Swiss International University (SIU) ORCID ID: 0009-0009-5070-1669 Submitted 21 February 2026; Revised 09 May 2026; Revised 22 June 2026; Accepted 02 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10005 Volume 3, December 2026, (SpecSDG10005) Abstract Global benchmarking instruments now shape how universities present, organise, and fund their sustainability work, yet their consequences for water-focused research remain poorly understood. This article develops an integrative review of three literatures that have grown largely in isolation: the sociology and scientometrics of university rankings, sustainability assessment in higher education, and research on water sustainability and Sustainable Development Goal 6. Drawing on peer-reviewed studies published mainly between 2003 and 2022, together with the published methodologies of the Times Higher Education Impact Rankings, UI GreenMetric, and the Sustainability Tracking, Assessment and Rating System, the review examines how water is represented in benchmarking architectures, through which institutional mechanisms participation can plausibly strengthen water research capacity, and where benchmarking threatens to distort that capacity. The analysis shows that water occupies a structurally weak position in most instruments: it is absent from academic world rankings, optional in the Impact Rankings, and folded into campus operations elsewhere. A conceptual framework is proposed that links benchmarking participation to water research capacity through three mechanisms, namely signalling and resource mobilisation, measurement and data infrastructure, and network formation, each conditioned by indicator validity and by reactive institutional behaviour. The framework yields testable propositions and design principles for ranking stewards, university leaders, and funders who want benchmarking to serve, rather than substitute for, substantive water scholarship. Keywords: university rankings, water sustainability, sustainable development goals, higher education, research assessment, benchmarking, SDG 6 1. Introduction Freshwater systems are under documented and intensifying pressure. Vörösmarty et al. (2010) mapped concurrent threats to human water security and river biodiversity at the global scale, and Mekonnen and Hoekstra (2016) estimated that four billion people face severe water scarcity. Hoekstra and Mekonnen (2012) traced the water footprint of humanity across production and consumption chains, and Gleick (2003) argued that twenty-first-century water challenges demand demand-side, efficiency-oriented solutions rather than supply expansion alone. These conditions define a research agenda that universities are uniquely positioned to advance: they train water professionals, operate campuses with measurable water demands, and produce the hydrological, engineering, and governance scholarship on which Sustainable Development Goal 6 (SDG 6, clean water and sanitation) depends. At the same time, universities increasingly organise their sustainability activity around external benchmarking instruments. The Times Higher Education (THE) Impact Rankings score institutions against the seventeen SDGs, including a dedicated SDG 6 pillar (Times Higher Education, 2024b). UI GreenMetric ranks campuses on criteria that include water usage (UI GreenMetric, n.d.). The Sustainability Tracking, Assessment and Rating System (STARS) awards credits for water performance within its operations category (Association for the Advancement of Sustainability in Higher Education [AASHE], n.d.). Participation in such systems has become a strategic act: it signals commitment to funders and applicants, structures internal reporting, and, as a long line of scholarship on rankings shows, changes the behaviour of the organisations being measured (Espeland & Sauder, 2007). The difficulty is that these two developments are studied by communities that rarely meet. Scholars of rankings have produced a rich critique of indicator validity and organisational reactivity (Gadd, 2021; Marginson, 2014; Selten et al., 2020), but they seldom examine environmental content, and almost never water specifically. Scholars of sustainability in higher education have compared assessment tools and documented implementation gaps (Alghamdi et al., 2017; Galleli et al., 2022; Lozano et al., 2015), but they treat water as one operational category among many rather than as a research field with its own capacity requirements. Water scholars, for their part, have mapped SDG-related publication trends (Salvia et al., 2019; Sianes et al., 2022) without asking how benchmarking participation feeds back into the research agendas they measure. The result is a precise and consequential gap: there is no integrated account of how global university benchmarking represents water sustainability, through which mechanisms participation could strengthen or weaken water research capacity, and what design features would make the difference. The gap matters for ranking stewards deciding how to weight and verify water indicators, for university leaders deciding whether SDG 6 participation is worth its reporting cost, and for funders and water agencies deciding whether ranking positions carry any signal about research capability. This article addresses that gap through an integrative review with a conceptual contribution. Following Torraco (2016) and Whittemore and Knafl (2005), it synthesises heterogeneous literatures in order to generate a new framework rather than to aggregate effect sizes. Three research questions structure the analysis: · RQ1. How do global university benchmarking systems represent water sustainability and management in their indicator architectures, and with what validity? · RQ2. Through which institutional mechanisms can benchmarking participation plausibly shape universities' water-related research capacity and agendas? · RQ3. Which unintended consequences of benchmarking threaten water-sustainability research, and what design principles follow for instrument stewards and universities? The contribution is threefold. First, the review provides the first systematic comparison of how four families of benchmarking instruments position water (RQ1), synthesised in Table 1. Second, it develops a conceptual framework, presented in Figure 1, that specifies three mechanisms linking benchmarking participation to water research capacity together with the conditions under which each operates (RQ2). Third, it derives risks and design principles that convert the rankings-critique literature into actionable guidance for the water case (RQ3). The remainder of the article reviews the relevant literatures, describes the review method, presents the findings organised around the three questions, and discusses implications, limitations, and the extent to which the identified gap has been closed. 2. Literature Review and Theoretical Background 2.1 Water Sustainability as a Research Imperative The empirical case for sustained water scholarship is not in dispute. Global assessments document concurrent threats to human water security and to river ecosystems (Vörösmarty et al., 2010), severe scarcity affecting four billion people (Mekonnen & Hoekstra, 2016), and consumption patterns whose water footprint extends far beyond the point of use (Hoekstra & Mekonnen, 2012). Gleick (2003) reframed the policy problem as one of managing demand, improving productivity per unit of water, and matching water quality to use, an agenda that requires interdisciplinary research spanning engineering, economics, and governance. Universities contribute to this agenda in two distinct capacities that the literature tends to conflate: as producers of water research and as water-using organisations. Marinho et al. (2014) illustrate the second capacity, documenting a water conservation programme at a Brazilian public university as a support for wider sustainable practice. The distinction matters for what follows, because benchmarking instruments differ precisely in which of the two capacities they measure. It also has a temporal dimension. Campus water performance can be improved within a budget cycle through metering, retrofits, and behavioural programmes, whereas water research capacity, in the form of laboratories, doctoral pipelines, long-term monitoring sites, and relationships with basin authorities, accumulates over decades and erodes quickly when funding signals turn elsewhere. Any instrument that measures the first capacity while purporting to speak for the second therefore risks rewarding fast, visible improvements at the expense of slow, structural ones. This asymmetry recurs throughout the analysis that follows. 2.2 Global Rankings: Measurement, Validity, and Reactivity Research on university rankings offers the theoretical core for this review. Espeland and Sauder (2007) established that public measures are reactive: organisations reshape themselves around the measure, so that rankings do not merely describe universities but recreate them. Subsequent scientometric work has specified what the major instruments actually capture. Selten et al. (2020) found that the Academic Ranking of World Universities, THE World University Rankings, and QS rankings are stable over time and that their indicators load principally on two latent factors, institutional reputation and research performance, raising the concern that the variables do not capture the broader quality concepts they claim to measure. Moed (2017) reached a compatible conclusion through a comparative analysis of five world rankings. Vernon et al. (2018), in a systematic review of ranking systems, identified twenty-four systems, evaluated thirteen against basic quality criteria, and reported that no generally accepted indicators exist and that no single system comprehensively evaluates research quality; among systems that disclose weights, the large majority of the weighting rewarded research or teaching quality, often via reputation surveys. Marginson (2014) questioned the social-science validity of ranking constructs, and Gadd (2021) argued that rankings reward institutions that grow up the tables rather than institutions that mature against their own missions. Hicks et al. (2015) distilled the corrective position into principles for responsible metrics, insisting that quantitative indicators should support, not replace, expert judgment. None of this literature, however, asks what these dynamics imply for a specific thematic field such as water. That question requires joining the rankings critique to the sustainability assessment literature. 2.3 Sustainability Assessment in Higher Education A parallel literature examines how universities institutionalise sustainability and how dedicated instruments assess it. Lozano (2006) analysed the organisational barriers that sustainability initiatives must overcome, and Lozano et al. (2015) found, through a worldwide survey, that declarations of commitment outrun implementation. Findler et al. (2019) reviewed research from 2005 to 2017 and reoriented the field from what universities do toward how their activities affect society, the environment, and the economy. Leal Filho et al. (2019) assessed whether sustainability teaching keeps pace with the SDGs, and Purcell et al. (2019) proposed the university as a living laboratory in which campus operations become sites of research and learning. Within this field, assessment instruments have received focused attention. Suwartha and Sari (2013) evaluated the early UI GreenMetric ranking as a tool for green university development, and Lauder et al. (2015) subjected the same instrument to a critical review of its design as a global campus sustainability ranking. Alghamdi et al. (2017) compared the indicator sets of sustainability assessment tools across universities, showing substantial variation in coverage. Urbanski and Leal Filho (2015) analysed early institutional data submitted to STARS. Atici et al. (2021) examined empirically how GreenMetric participation relates to standing in world university rankings, connecting the campus-greening and academic-prestige literatures. Galleli et al. (2022) compared UI GreenMetric with the THE World University Rankings against the Berlin Principles and concluded that both exhibit structural differences and methodological limitations, so that institutions must choose instruments contextually rather than expect a single valid measure. For the SDG-specific instruments, De la Poza et al. (2021) used THE Impact Rankings data to model universities' SDG reporting, while Bautista-Puig et al. (2022), analysing the 2019 to 2021 editions, found severe methodological inconsistencies that in their assessment produce a distorted view of sustainability performance, alongside genuine reputational opportunities for less-prominent institutions. 2.4 SDG Measurement and the Research Agenda A third literature measures how the SDGs are reshaping research itself. Salvia et al. (2019) assessed SDG-related research trends and the alignment between local issues and global agendas. Sianes et al. (2022) traced the scientometric imprint of the SDGs on academic research. Armitage et al. (2020) demonstrated a foundational measurement problem: independent bibliometric approaches to identifying SDG-related publications showed little overlap, and the choice of approach altered country rankings, leading the authors to advise caution toward SDG rankings and tools. At the level of goal politics, Forestier and Kim (2020) documented selective prioritisation among the SDGs by national governments, and Heleta and Bagus (2021) argued that SDG frameworks neglect higher education capacity in low-income countries and can reinforce global inequality. 2.5 Synthesis and Gap Read together, these literatures supply all the raw material for an account of benchmarking and water research, yet none assembles it. The rankings literature explains reactivity and indicator invalidity but is thematically blind. The sustainability assessment literature evaluates instruments but concentrates on campus operations and aggregate SDG engagement rather than on any single goal's research base. The SDG measurement literature shows that classifying water-related research is itself unstable, which undermines the evidentiary foundation of any water indicator, but it does not follow the consequences into institutional behaviour. What the combined literature has not resolved is (a) a systematic description of where water sits in benchmarking architectures, (b) a specification of the mechanisms by which benchmarking participation could change water research capacity, and (c) an assessment of which known ranking pathologies bear most directly on water. Those three unresolved items correspond to RQ1 through RQ3 and define the contribution of this review. 3. Method 3.1 Review Design The study uses an integrative review design, which is appropriate when the goal is to synthesise conceptually heterogeneous literatures and to generate new frameworks rather than to estimate pooled effects (Torraco, 2016). Whittemore and Knafl (2005) specify the stages followed here: problem identification, literature search, data evaluation, data analysis, and presentation. The design is complemented by a documentary analysis of the published methodologies of three benchmarking instruments (Times Higher Education, 2024a, 2024b; UI GreenMetric, n.d.; AASHE, n.d.), because indicator architectures are primary documents that the peer-reviewed literature discusses but does not reproduce in full. 3.2 Search Strategy Searches were conducted in Scopus, Web of Science, and Google Scholar for literature published between January 2003 and August 2026, with the core corpus concentrated in 2013 to 2022. The lower bound admits seminal works on water policy and rankings sociology; the emphasis on the last decade reflects the founding of the THE Impact Rankings and the SDG era. Example search strings, adapted to each database's syntax, included: ("university ranking*" OR "league table*" OR benchmark*) AND (sustainab* OR "SDG*"); ("UI GreenMetric" OR "STARS" OR "Impact Rankings") AND (universit* OR "higher education"); (water OR "SDG 6" OR "water management") AND (universit* OR campus OR "higher education") AND (research OR ranking OR assessment); and ("research assessment" OR bibliometric*) AND ("sustainable development goals"). Reference lists of included articles were snowballed in both directions. 3.3 Inclusion and Exclusion Criteria Sources were included if they (a) were peer-reviewed journal articles, or methodology documents published by the steward of a benchmarking instrument; (b) addressed at least one of the three review domains, namely rankings and research assessment, sustainability assessment in higher education, or water sustainability and SDG measurement; and (c) reported an identifiable argument, conceptualisation, or empirical result relevant to the research questions. Sources were excluded if they (a) were editorials, theses, or conference abstracts without full analysis; (b) addressed campus sustainability without any assessment, benchmarking, or research-capacity dimension; or (c) could not be verified against their publisher's bibliographic record. Screening proceeded in two stages, title and abstract followed by full text, with the research questions as the screening rubric. Thirty-six sources satisfied all criteria: thirty-two peer-reviewed articles and four instrument methodology documents. 3.4 Analysis and Framework Derivation Included sources were coded against three analytic categories derived from the research questions: representation (how an instrument defines, weights, and verifies water content), mechanism (any process by which measurement is claimed or shown to change institutional behaviour or research activity), and pathology (any documented distortion attributable to measurement). Constant comparison across the three literatures generated the framework in Figure 1: mechanisms proposed in the rankings and sustainability literatures were retained only where at least two independent sources supported the underlying process, and each mechanism was then specified for the water case using the instrument documents. The framework is therefore a conceptual synthesis, and its water-specific pathways are stated as propositions to be tested, not as established findings. 3.5 Rigour and Trustworthiness Several safeguards address the known weaknesses of integrative reviews. Selection bias was limited by searching three databases, by snowballing, and by including critical as well as favourable evaluations of every instrument discussed. Verification bias was limited by checking every cited source's bibliographic record against the Crossref or OpenAlex registry and by consulting instrument methodologies in their steward's own publications rather than through secondary description. Interpretive claims are marked as such throughout, and single studies are attributed as single studies. Two scope boundaries should be stated plainly: the review covers documents published in English, which underrepresents scholarship from several water-stressed regions, and it does not attempt a quantitative meta-analysis, because the included studies do not share comparable outcome measures. No systematic-review reporting checklist was applied, and no claim is made about the exhaustiveness of the corpus beyond the stated search protocol. 4. Findings 4.1 Where Water Sits in Benchmarking Architectures (RQ1) The first finding is structural: across the four families of instruments that dominate global university benchmarking, water occupies positions of sharply different visibility, and in no case is water research capacity measured directly and verifiably. Table 1 summarises the comparison. Academic world rankings contain no water-specific indicators at all. Their variables reduce, empirically, to reputation and aggregate research performance (Selten et al., 2020), and systematic evaluation finds their indicator sets dominated by research and teaching prestige with no generally accepted standards (Vernon et al., 2018; Moed, 2017). Water research enters these instruments only as an undifferentiated contribution to publication and citation counts. A university could dismantle its entire water institute without any detectable movement in its world ranking position, a property that follows directly from the indicator structure documented in the scientometric literature. The THE Impact Rankings are the only global instrument with a dedicated water pillar. The SDG 6 methodology allocates 27 percent of the pillar score to research on clean water and sanitation, assessed through Scopus-based citation impact and publication volume for 2018 to 2022, with the remaining weight distributed across water consumption (19 percent), water usage and care (23 percent), water reuse (12 percent), and community engagement (19 percent) (Times Higher Education, 2024b). Two architectural features qualify this apparent prominence. First, participation is selective: universities submit data on as many SDGs as they choose, and the overall score combines SDG 17 with each institution's best three other goals (Times Higher Education, 2024a). SDG 6 therefore competes for attention with sixteen alternatives, and institutions rationally submit their strongest goals, a dynamic of goal-level selectivity that mirrors the cherry-picking Forestier and Kim (2020) documented among national governments. Second, the evidence base is largely self-provided, with missing data scored as zero (Times Higher Education, 2024a, 2024b), and content analysis of the 2019 to 2021 editions found severe methodological inconsistencies that distort the resulting picture of sustainability performance (Bautista-Puig et al., 2022). The campus-greening instruments treat water as operations. UI GreenMetric includes water usage among its six weighted criteria, alongside setting and infrastructure, energy and climate change, waste, transportation, and education and research (UI GreenMetric, n.d.), on the basis of self-reported campus data whose design has been critically reviewed since the instrument's early years (Lauder et al., 2015; Suwartha & Sari, 2013). STARS awards water-related credits within its operations category in a transparent, self-reporting, points-based framework (AASHE, n.d.; Urbanski & Leal Filho, 2015). In both instruments, water performance means campus water performance; research on water appears, if at all, inside generic education and research credits. The answer to RQ1 is therefore that benchmarking architectures represent water either not at all, or optionally, or operationally. Only one instrument measures water research, it does so for a self-selected subset of institutions, and it relies on bibliometric classification of water-related publications, a procedure that Armitage et al. (2020) showed to be unstable, since independent SDG mapping approaches produced little overlap and materially different rankings. The validity qualifier in RQ1 is thus answered in the negative: no current instrument offers a valid, comparable measure of water research capacity across institutions. A comparison across the four architectures also reveals a trade-off between visibility and verifiability. The instruments that make water most visible, the SDG 6 pillar and the campus-greening rankings, rest mainly on self-reported evidence, while the instruments with the most externally verifiable data, the bibliometrics-driven world rankings, make water invisible. The comparative literature reaches a consistent conclusion about this situation: Galleli et al. (2022) found that neither a campus-greening instrument nor an academic ranking satisfies the Berlin Principles fully and advised contextual selection rather than reliance on any single system, and Alghamdi et al. (2017) documented wide variation in which indicators sustainability assessment tools include at all. For water specifically, this means that an institution seeking an external mirror for its water performance must triangulate at least two instruments with different blind spots, and that any single-instrument account of a university's water standing should be treated as partial by construction. Table 1 Water Sustainability in Four Families of Global University Benchmarking Instruments Instrument family (steward) Primary orientation Placement of water Water-related indicators and data basis Key sources Academic world rankings (ARWU, THE WUR, QS) Reputation and aggregate research performance Absent; no dedicated environmental or water indicators Water research counted only inside aggregate publication and citation measures; bibliometric data and reputation surveys Moed (2017); Selten et al. (2020); Vernon et al. (2018) THE Impact Rankings (Times Higher Education) Contribution to the 17 SDGs; overall score combines SDG 17 with each institution's best three other goals Dedicated, optional SDG 6 pillar Research on clean water and sanitation (27%); water consumption (19%); water usage and care (23%); water reuse (12%); water in the community (19%); Scopus data plus self-submitted evidence, with missing data scored zero Times Higher Education (2024a, 2024b); Bautista-Puig et al. (2022); De la Poza et al. (2021) UI GreenMetric (Universitas Indonesia) Campus greening and infrastructure Water usage as one of six weighted criteria Campus water usage reported by institutions; weightings under continuous review UI GreenMetric (n.d.); Lauder et al. (2015); Suwartha and Sari (2013) STARS (AASHE) Institutional sustainability self-assessment Water credits within the Operations category Points-based credits including water use; transparent, voluntary self-reporting AASHE (n.d.); Urbanski and Leal Filho (2015) Note. Indicator names, weights, and category placements are taken from the instruments' published methodologies as cited in the final column; characterisations of the academic world rankings reflect the scientometric analyses cited. ARWU = Academic Ranking of World Universities; THE WUR = Times Higher Education World University Rankings; QS = Quacquarelli Symonds; SDG = Sustainable Development Goal; STARS = Sustainability Tracking, Assessment and Rating System; AASHE = Association for the Advancement of Sustainability in Higher Education. 4.2 Mechanisms Linking Benchmarking Participation to Water Research Capacity (RQ2) The second finding is that, despite these representational weaknesses, the literature supports three distinct mechanisms through which benchmarking participation can plausibly build water research capacity. Figure 1 assembles them into a conceptual framework; each pathway is stated here as a proposition grounded in the sources that support the underlying process. The first mechanism is signalling and resource mobilisation. Rankings are reactive instruments: organisations reallocate attention and resources toward what is measured (Espeland & Sauder, 2007), and universities have been shown to convert sustainability commitments into structures and budgets unevenly, with visible external commitments outrunning implementation (Lozano et al., 2015). Where an institution elects the SDG 6 pillar, the 27 percent research weighting (Times Higher Education, 2024b) creates, for the first time in any global instrument, a direct reputational return on water scholarship. Bautista-Puig et al. (2022) found that the Impact Rankings offer reputational opportunities precisely to institutions outside the traditional elite, which suggests that the signalling mechanism may operate most strongly for universities in water-stressed middle-income regions, where world rankings offer them little. Proposition 1: universities that elect SDG 6 will, other conditions equal, increase internal allocation to water research relative to observationally similar non-electing institutions. The second mechanism is measurement and data infrastructure. Benchmarking obliges institutions to meter, audit, and document their own water systems: consumption per capita, wastewater treatment, reuse policies (Times Higher Education, 2024b), and the operational categories of GreenMetric and STARS (UI GreenMetric, n.d.; AASHE, n.d.). This reporting burden creates campus water data that did not previously exist in comparable form, and campus data are the raw material of the living-laboratory model in which operations become research sites (Purcell et al., 2019). The case documented by Marinho et al. (2014), where a university water conservation programme supported wider sustainable practice, illustrates the pathway from operational measurement to applied scholarship. This mechanism reframes the two university capacities distinguished in Section 2.1: benchmarking of the university as water user can subsidise the university as water researcher. Proposition 2: institutions with sustained participation in operations-focused instruments will produce more campus-based water research than non-participants. The third mechanism is network formation and agenda alignment. SDG-structured benchmarking embeds universities in a shared classification of societal problems, which lowers the cost of identifying partners, and the community components of the SDG 6 pillar explicitly reward off-campus cooperation on water security (Times Higher Education, 2024b). The literature on universities and the SDGs argues that such engagement reorients institutional missions toward societal impact (Findler et al., 2019; Leal Filho et al., 2019), and bibliometric evidence confirms that the SDG framework has left a measurable imprint on research agendas (Sianes et al., 2022; Salvia et al., 2019). Proposition 3: benchmarking participation increases the share of water research conducted with non-academic partners and oriented to local water problems. The three mechanisms are not independent, and their interactions carry analytical weight. Signalling without measurement produces commitments that outrun implementation, the pattern Lozano et al. (2015) observed across the sustainability declarations of the preceding two decades, because reputational incentives arrive before the data systems needed to act on them. Measurement without signalling produces data that remain administrative, since without reputational or funding stakes there is little pull to convert campus water records into research questions. Network formation amplifies both: partners demand data, which strengthens the measurement pathway, and partnerships generate the demonstrable community engagement that the SDG 6 pillar rewards (Times Higher Education, 2024b), which strengthens signalling. A capacity-building account of benchmarking therefore predicts the strongest effects where all three mechanisms operate together, typically in institutions that elect SDG 6, sustain operational reporting, and hold standing relationships with water authorities. These mechanisms answer RQ2, but the framework in Figure 1 also specifies their conditions. Each pathway passes through two moderating filters: indicator validity, which is currently weak (Armitage et al., 2020; Galleli et al., 2022), and institutional response type, which ranges from substantive investment to symbolic compliance. The next subsection examines the second filter. Figure 1. Conceptual framework linking benchmarking participation to water sustainability and management research capacity through three mechanisms (M1 to M3), conditioned by indicator validity and by institutional response. Source: author's elaboration from the reviewed literature. 4.3 Risks: How Benchmarking Can Distort Water Research (RQ3) The third finding is that every major pathology documented in the rankings literature has a specific and foreseeable water-sector expression. Reactivity can become gaming. Espeland and Sauder (2007) showed that measured organisations manage the measure, not only the underlying performance. In the water case, the combination of self-provided evidence and zero-scoring of missing data (Times Higher Education, 2024a) rewards documentation capacity as much as water performance, and the inconsistencies identified by Bautista-Puig et al. (2022) indicate that the verification layer is not yet strong enough to separate the two. Institutions with professional rankings offices can therefore outperform institutions with stronger water science but weaker reporting, a concern consistent with the finding of Heleta and Bagus (2021) that SDG frameworks disadvantage under-resourced institutions in low-income countries, including many in the most water-stressed regions. Classification instability can misdirect credit. The research component of the SDG 6 score depends on bibliometric identification of water-related publications, yet Armitage et al. (2020) found little overlap between independent approaches to exactly this task and showed that the choice of approach changes rankings. Until SDG 6 publication mapping stabilises, research-weighted water scores contain an unquantified layer of classification noise, and universities optimising against them may be optimising against an artefact. Goal selectivity can hollow out the signal. Because institutions submit their strongest goals (Times Higher Education, 2024a), the population ranked on SDG 6 is self-selected, so pillar positions cannot be read as a census of global water research capacity. Forestier and Kim (2020) showed that selective SDG prioritisation at the national level carries governance costs; the same logic implies that water, a goal requiring expensive infrastructure and specialised research capacity, risks systematic under-election relative to goals that most institutions can document cheaply. This is an interpretive extension of their finding, offered here as a proposition rather than an established result. Metric fixation can displace judgment. The general corrective is well established: indicators should support expert judgment, not replace it (Hicks et al., 2015), and instruments reward growth up the table rather than maturity against mission (Gadd, 2021). For water, the mission-relevant questions, such as whether research addresses the basin problems of the university's own region, are precisely the ones that Salvia et al. (2019) found imperfectly aligned between local issues and global agendas, and no current indicator captures them. 4.4 Design Principles Answering the second half of RQ3, four design principles follow from the analysis, each traceable to the evidence above. First, verify before weighting: research-heavy water scores should not exceed the reliability of SDG 6 publication mapping, which argues for published sensitivity analyses across mapping approaches (Armitage et al., 2020). Second, reward disclosure symmetry: instruments should distinguish absent performance from absent documentation, since zero-scoring missing data conflates the two (Times Higher Education, 2024a) and penalises under-resourced institutions (Heleta & Bagus, 2021). Third, connect the operational and research ledgers: instruments already collect campus water data; publishing them in reusable form would let benchmarking directly subsidise living-laboratory research (Purcell et al., 2019; Marinho et al., 2014). Fourth, benchmark contextually: following the conclusion of Galleli et al. (2022) that no single instrument is best, universities should select and interpret instruments against their own water context, and evaluators should follow the principle that metrics inform rather than decide (Hicks et al., 2015). 5. Discussion 5.1 Theoretical Implications The review's central theoretical claim is that reactivity theory (Espeland & Sauder, 2007) gains explanatory power when it is made goal-specific. Applied at the level of whole institutions, reactivity predicts generic ranking-seeking behaviour. Applied at the level of a single SDG, it predicts differentiated behaviour: election or avoidance of the goal, substantive or symbolic response, and reallocation across goals as relative prices change. The framework in Figure 1 formalises this by treating benchmarking participation as an institutional choice whose research consequences run through three mechanisms and two filters. This specification also connects reactivity theory to the sustainability implementation literature: the gap between declaration and implementation that Lozano et al. (2015) documented is, in the framework's terms, the symbolic branch of the institutional response filter. For the scientometrics of the SDGs, the analysis converts the measurement instability shown by Armitage et al. (2020) from a technical caveat into a theoretical variable, since classification noise determines how much of the signalling mechanism reaches actual water research rather than an artefact of mapping. 5.2 Practical and Policy Implications For ranking stewards, the analysis implies that the credibility of water pillars now depends less on additional indicators than on verification and on published sensitivity of research scores to mapping choices. For university leaders, the framework offers a decision structure: SDG 6 election is most defensible where the institution can pair reporting with substantive investment, and the measurement mechanism means that even operations-focused participation can be converted into research assets if campus water data are treated as research infrastructure. For funders and water agencies, the self-selected nature of SDG 6 pillar populations means ranking positions should not be used as a screen for research capability; the bibliometric record and expert review remain the appropriate instruments, used under responsible-metrics principles (Hicks et al., 2015). For policymakers in water-stressed regions, the equity findings counsel support for reporting capacity, since otherwise benchmarking will systematically understate the water work of the institutions closest to the problem (Heleta & Bagus, 2021). There is also a positive policy reading of the analysis. Because the measurement mechanism runs through data that universities must collect anyway, national water agencies could treat benchmarked campuses as a distributed observation network: standardised consumption, treatment, and reuse data across hundreds of institutions constitute evidence about demand-side water management of exactly the kind the soft-path agenda requires (Gleick, 2003). Realising that value requires only that instrument stewards publish operational water data in reusable form, a change that costs participants nothing beyond what current reporting already demands. 5.3 Limitations The limitations follow from the method. First, this is an integrative review: the framework is a conceptual synthesis, its propositions are untested, and no causal claim about benchmarking and water research output is made or warranted. Second, the corpus is English-language and concentrated in journals indexed by the major databases, which underrepresents regions where water stress is most acute and where the equity effects discussed above matter most. Third, the documentary analysis rests on instrument methodologies as published in 2024 editions and current technical manuals; benchmarking methodologies change frequently, and the specific weights cited here will date. Fourth, the review depends in places on single studies, notably for the content analysis of Impact Rankings inconsistencies (Bautista-Puig et al., 2022) and for SDG mapping instability (Armitage et al., 2020); these are attributed as single studies, and the framework would need revision if replication fails. Fifth, no formal quality scoring of included studies was undertaken beyond the stated inclusion criteria, which is a recognised trade-off of integrative designs (Whittemore & Knafl, 2005). 5.4 Future Research Each proposition in Section 4.2 defines an empirical study. Proposition 1 invites a difference-in-differences design comparing water research investment in SDG 6-electing and non-electing universities, feasible once panel data on pillar participation accumulate. Proposition 2 invites bibliometric analysis of campus-based water research among long-run GreenMetric and STARS participants. Proposition 3 invites co-authorship and funding-acknowledgement analysis of partnered water research before and after benchmarking entry. Beyond the propositions, two measurement studies are prerequisite to all evaluative work: a water-specific replication of the mapping comparison of Armitage et al. (2020) confined to SDG 6 queries, and an audit study of the verification practices behind self-reported water evidence. Finally, qualitative work inside universities is needed to observe the response filter directly, distinguishing substantive from symbolic SDG 6 engagement in the tradition of implementation research (Lozano et al., 2015). 5.5 How Far the Gap Was Closed Of the three unresolved items identified in Section 2.5, the first, the representational question, is now closed to the extent that public methodologies allow: Table 1 provides the systematic comparison that the literature lacked. The second, the mechanism question, is closed at the conceptual level: the framework specifies pathways and conditions, but their empirical weight remains unmeasured. The third, the pathology question, is closed as translation: known ranking distortions have been given specific water-sector expressions and countermeasures, though several of these expressions are propositions rather than observations. The review therefore converts an unstructured gap into a structured research programme; it does not, and by design cannot, supply the causal evidence that programme calls for. 6. Conclusion Global university benchmarking has begun to measure water, but it measures it unevenly: not at all in the academic world rankings, optionally and with weak verification in the SDG-based rankings, and operationally in the campus-greening instruments. This integrative review joined the rankings, sustainability assessment, and water research literatures to show what follows from that architecture. Benchmarking participation can strengthen water sustainability and management research through signalling that attaches reputational value to water scholarship, through measurement that creates campus water data usable as research infrastructure, and through networks that align research with local water problems. The same participation can weaken the field through gaming, classification noise, goal selectivity, and metric fixation, and these risks fall hardest on institutions in water-stressed, resource-poor settings. Whether benchmarking advances or distorts water research is therefore not a property of benchmarking as such but of indicator validity and institutional response, the two filters at the centre of the framework proposed here. The propositions derived from that framework set the empirical agenda; the design principles indicate what stewards and universities can change without waiting for it. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics. This study is a review of published literature and publicly available documents; it involved no human participants, animals, or personal data, and no ethical approval was required. Data Availability. No new data were created or analysed in this study. All sources synthesised are cited and publicly available through the references listed. References Alghamdi, N., den Heijer, A., & de Jonge, H. (2017). Assessment tools' indicators for sustainability in universities: An analytical overview. International Journal of Sustainability in Higher Education, 18(1), 84–115. https://doi.org/10.1108/IJSHE-04-2015-0071 Armitage, C. S., Lorenz, M., & Mikki, S. (2020). Mapping scholarly publications related to the Sustainable Development Goals: Do independent bibliometric approaches get the same results? Quantitative Science Studies, 1(3), 1092–1108. https://doi.org/10.1162/qss_a_00071 Association for the Advancement of Sustainability in Higher Education. (n.d.). STARS technical manual. Retrieved August 8, 2026, from https://stars.aashe.org/pages/about/technical-manual.html Atici, K. B., Yasayacak, G., Yildiz, Y., & Ulucan, A. (2021). Green University and academic performance: An empirical study on UI GreenMetric and World University Rankings. Journal of Cleaner Production, 291, 125289. https://doi.org/10.1016/j.jclepro.2020.125289 Bautista-Puig, N., Orduña-Malea, E., & Pérez-Esparrells, C. (2022). Enhancing sustainable development goals or promoting universities? An analysis of the Times Higher Education Impact Rankings. International Journal of Sustainability in Higher Education, 23(8), 211–231. https://doi.org/10.1108/IJSHE-07-2021-0309 De la Poza, E., Merello, P., Barberá, A., & Celani, A. (2021). Universities' reporting on SDGs: Using THE Impact Rankings to model and measure their contribution to sustainability. Sustainability, 13(4), 2038. https://doi.org/10.3390/su13042038 Espeland, W. N., & Sauder, M. (2007). Rankings and reactivity: How public measures recreate social worlds. American Journal of Sociology, 113(1), 1–40. https://doi.org/10.1086/517897 Findler, F., Schönherr, N., Lozano, R., Reider, D., & Martinuzzi, A. (2019). The impacts of higher education institutions on sustainable development: A review and conceptualization. International Journal of Sustainability in Higher Education, 20(1), 23–38. https://doi.org/10.1108/IJSHE-07-2017-0114 Forestier, O., & Kim, R. E. (2020). Cherry-picking the Sustainable Development Goals: Goal prioritization by national governments and implications for global governance. Sustainable Development, 28(5), 1269–1278. https://doi.org/10.1002/sd.2082 Gadd, E. (2021). Mis-measuring our universities: Why global university rankings don't add up. Frontiers in Research Metrics and Analytics, 6, 680023. https://doi.org/10.3389/frma.2021.680023 Galleli, B., Teles, N. E. B., Santos, J. A. R. dos, Freitas-Martins, M. S., & Hourneaux Junior, F. (2022). Sustainability university rankings: A comparative analysis of UI GreenMetric and the Times Higher Education World University Rankings. International Journal of Sustainability in Higher Education, 23(2), 404–425. https://doi.org/10.1108/IJSHE-12-2020-0475 Gleick, P. H. (2003). Global freshwater resources: Soft-path solutions for the 21st century. Science, 302(5650), 1524–1528. https://doi.org/10.1126/science.1089967 Heleta, S., & Bagus, T. (2021). Sustainable development goals and higher education: Leaving many behind. Higher Education, 81(1), 163–177. https://doi.org/10.1007/s10734-020-00573-8 Hicks, D., Wouters, P., Waltman, L., de Rijcke, S., & Rafols, I. (2015). Bibliometrics: The Leiden Manifesto for research metrics. Nature, 520(7548), 429–431. https://doi.org/10.1038/520429a Hoekstra, A. Y., & Mekonnen, M. M. (2012). The water footprint of humanity. Proceedings of the National Academy of Sciences, 109(9), 3232–3237. https://doi.org/10.1073/pnas.1109936109 Lauder, A., Sari, R. F., Suwartha, N., & Tjahjono, G. (2015). Critical review of a global campus sustainability ranking: GreenMetric. Journal of Cleaner Production, 108, 852–863. https://doi.org/10.1016/j.jclepro.2015.02.080 Leal Filho, W., Shiel, C., Paço, A., Mifsud, M., Ávila, L. V., Brandli, L. L., Molthan-Hill, P., Pace, P., Azeiteiro, U. M., Vargas, V. R., & Caeiro, S. (2019). Sustainable Development Goals and sustainability teaching at universities: Falling behind or getting ahead of the pack? Journal of Cleaner Production, 232, 285–294. https://doi.org/10.1016/j.jclepro.2019.05.309 Lozano, R. (2006). Incorporation and institutionalization of SD into universities: Breaking through barriers to change. Journal of Cleaner Production, 14(9–11), 787–796. https://doi.org/10.1016/j.jclepro.2005.12.010 Lozano, R., Ceulemans, K., Alonso-Almeida, M., Huisingh, D., Lozano, F. J., Waas, T., Lambrechts, W., Lukman, R., & Hugé, J. (2015). A review of commitment and implementation of sustainable development in higher education: Results from a worldwide survey. Journal of Cleaner Production, 108, 1–18. https://doi.org/10.1016/j.jclepro.2014.09.048 Marginson, S. (2014). University rankings and social science. European Journal of Education, 49(1), 45–59. https://doi.org/10.1111/ejed.12061 Marinho, M., Gonçalves, M. do S., & Kiperstok, A. (2014). Water conservation as a tool to support sustainable practices in a Brazilian public university. Journal of Cleaner Production, 62, 98–106. https://doi.org/10.1016/j.jclepro.2013.06.053 Mekonnen, M. M., & Hoekstra, A. Y. (2016). Four billion people facing severe water scarcity. Science Advances, 2(2), e1500323. https://doi.org/10.1126/sciadv.1500323 Moed, H. F. (2017). A critical comparative analysis of five world university rankings. Scientometrics, 110(2), 967–990. https://doi.org/10.1007/s11192-016-2212-y Purcell, W. M., Henriksen, H., & Spengler, J. D. (2019). Universities as the engine of transformational sustainability toward delivering the sustainable development goals: "Living labs" for sustainability. International Journal of Sustainability in Higher Education, 20(8), 1343–1357. https://doi.org/10.1108/IJSHE-02-2019-0103 Salvia, A. L., Leal Filho, W., Brandli, L. L., & Griebeler, J. S. (2019). Assessing research trends related to Sustainable Development Goals: Local and global issues. Journal of Cleaner Production, 208, 841–849. https://doi.org/10.1016/j.jclepro.2018.09.242 Selten, F., Neylon, C., Huang, C.-K., & Groth, P. (2020). A longitudinal analysis of university rankings. Quantitative Science Studies, 1(3), 1109–1135. https://doi.org/10.1162/qss_a_00052 Sianes, A., Vega-Muñoz, A., Tirado-Valencia, P., & Ariza-Montes, A. (2022). Impact of the Sustainable Development Goals on the academic research agenda: A scientometric analysis. PLOS ONE, 17(3), e0265409. https://doi.org/10.1371/journal.pone.0265409 Suwartha, N., & Sari, R. F. (2013). Evaluating UI GreenMetric as a tool to support green universities development: Assessment of the year 2011 ranking. Journal of Cleaner Production, 61, 46–53. https://doi.org/10.1016/j.jclepro.2013.02.034 Times Higher Education. (2024a). Impact Rankings 2024: Methodology. Retrieved August 8, 2026, from https://www.timeshighereducation.com/world-university-rankings/impact-rankings-2024-methodology Times Higher Education. (2024b). Impact Rankings 2024: Clean water and sanitation (SDG 6) methodology. Retrieved August 8, 2026, from https://www.timeshighereducation.com/impact-rankings-2024-clean-water-and-sanitation-sdg-6-methodology Torraco, R. J. (2016). Writing integrative literature reviews: Using the past and present to explore the future. Human Resource Development Review, 15(4), 404–428. https://doi.org/10.1177/1534484316671606 UI GreenMetric. (n.d.). Methodology. Universitas Indonesia. Retrieved August 8, 2026, from https://greenmetric.ui.ac.id/about/methodology Urbanski, M., & Leal Filho, W. (2015). Measuring sustainability at universities by means of the Sustainability Tracking, Assessment and Rating System (STARS): Early findings from STARS data. Environment, Development and Sustainability, 17(2), 209–220. https://doi.org/10.1007/s10668-014-9564-3 Vernon, M. M., Balas, E. A., & Momani, S. (2018). Are university rankings useful to improve research? A systematic review. PLOS ONE, 13(3), e0193762. https://doi.org/10.1371/journal.pone.0193762 Vörösmarty, C. J., McIntyre, P. B., Gessner, M. O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S. E., Sullivan, C. A., Reidy Liermann, C., & Davies, P. M. (2010). Global threats to human water security and river biodiversity. Nature, 467(7315), 555–561. https://doi.org/10.1038/nature09440 Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Hashtags: #SDG6 #CleanWaterAndSanitation #SDG4 #QualityEducation #WaterSustainability #UniversityRankings #GlobalBenchmarking #HigherEducation #ResearchAssessment #CampusSustainability #U7YJournal #U7Y #AcademicResearch

Author: Liam Johnson Affiliation: Swiss International University (SIU) ORCID ID: 0009-0005-3168-1739 Submitted 17 April 2026; Revised 13 June 2026; Accepted 21 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10006 Volume 3, December 2026, (SpecSDG10006) Abstract Agritech innovation increasingly depends on partnerships that connect universities with firms, governments, advisory services, and farming communities, yet scholarship on these partnerships is split between two literatures that rarely engage each other. Research on university–industry collaboration offers mature accounts of engagement channels, barriers, and governance, but it synthesizes evidence across sectors and seldom treats agriculture as a distinct institutional setting. Social science on digital agriculture and agricultural innovation systems examines farms, advisory networks, and food system transitions, but it does not theorize the organizational forms through which universities enter these systems. This article reports an integrative review that connects the two bodies of work. Drawing on peer-reviewed studies published mainly between 2016 and 2026, identified through structured searches of Scopus, Web of Science, and Google Scholar and screened against explicit inclusion criteria, the review pursues three questions: which partnership configurations link universities to agritech innovation, through which mechanisms those configurations translate academic research into innovation outcomes, and under which contextual conditions the mechanisms operate in high-income and in low- and middle-income settings. The synthesis produces a typology of five partnership configurations and a conceptual framework that links configurations, translation mechanisms, contextual conditions, and the directionality of innovation. The central argument is that partnership form must match the knowledge mechanism it is intended to carry, and that context and responsibility considerations condition every pathway from academic research to agritech outcomes. Implications for university leaders, policymakers, and comparative research are set out. Keywords: agritech innovation, university–industry collaboration, technology transfer, digital agriculture, agricultural innovation systems, integrative review 1. Introduction Agricultural technology has become one of the most active frontiers of innovation policy. Digital sensing, data analytics, automation, and precision management promise gains in productivity and sustainability at the same time (Basso & Antle, 2020; Wolfert et al., 2017), and perspective pieces in the food systems literature argue that a broad portfolio of such innovations can accelerate the transition towards sustainable food systems (Herrero et al., 2020). The promise, however, is unevenly realized. Lowenberg-DeBoer and Erickson (2019) found that satellite-based guidance and associated automation spread as quickly as any major agricultural technology in history, while adoption of variable-rate applications, which demand more knowledge-intensive integration, rarely exceeded twenty percent of farms. In low- and middle-income countries, reviews of digital advisory services report substantial potential alongside persistent difficulties of scale and financing (Aker, 2011; Deichmann et al., 2016; Fabregas et al., 2019). The pattern that emerges from this evidence is not a shortage of technology but a shortage of arrangements that connect research to heterogeneous users under real institutional conditions. Universities sit at the center of this connection problem. They conduct much of the basic and applied research on which agritech draws, they train the agronomists, engineers, and data scientists who staff the sector, and they are increasingly expected to contribute directly to innovation as part of a third mission that extends beyond teaching and research (Compagnucci & Spigarelli, 2020). How universities organize that contribution is the subject of a large scholarship on university–industry collaboration, which has mapped engagement channels (D'Este & Patel, 2007), distinguished relational engagement from commercialization (Perkmann et al., 2013), synthesized organizational forms and process factors (Ankrah & Al-Tabbaa, 2015), and tracked the field's growth through bibliometric and systematic reviews (Mascarenhas et al., 2018; Perkmann et al., 2021; Skute et al., 2019). The difficulty is that these two research conversations have developed with little reference to each other, and the space between them is precisely where agritech partnerships operate. The university–industry collaboration literature is deliberately cross-sectoral: its major reviews synthesize evidence spanning many industries and do not treat agriculture as a distinct analytic category with its own user structures, advisory institutions, and public-good characteristics (Ankrah & Al-Tabbaa, 2015; Perkmann et al., 2013, 2021; Skute et al., 2019). The agricultural literature, for its part, has developed sophisticated accounts of innovation systems, digitalization, and advisory networks (Fielke et al., 2020; Ingram & Maye, 2020; Klerkx et al., 2019; World Bank, 2012), but in these accounts universities typically appear as one actor among many; the partnership configurations through which they engage, and the mechanisms those configurations carry, are not themselves theorized. The consequence, and the gap this review addresses, is that neither literature can currently answer a question that matters to university leaders designing agritech programmes, to ministries and funders deciding which partnership instruments to support, and to researchers seeking cumulative knowledge: which forms of university partnership advance agritech innovation, through which mechanisms, and under which conditions. This article answers that question through an integrative review with a conceptual contribution. Following the methodological tradition of Torraco (2005) and Whittemore and Knafl (2005), it reads the two literatures against each other and synthesizes them into a typology and a framework. Three research questions structure the review: RQ1. Through which partnership configurations do universities engage with industry, government, advisory services, and farming communities in agritech innovation? RQ2. Through which mechanisms do these configurations translate academic research into agritech innovation outcomes? RQ3. Under which contextual conditions do these mechanisms operate, and how do those conditions differ between high-income and low- and middle-income settings? The contribution is threefold. First, the review consolidates dispersed evidence into a typology of five partnership configurations observable in agritech (Table 1). Second, it proposes a conceptual framework, presented in Figure 1, that links configurations to innovation outcomes through identifiable translation mechanisms, moderated by contextual conditions and steered by considerations of directionality and responsibility. Third, it derives from this framework a research agenda for comparative work on university partnerships in agriculture. The remainder of the article proceeds as follows: the next section reviews the two parent literatures thematically; the Method section sets out the review protocol; the Findings section develops the typology, mechanisms, and conditions; the Discussion draws theoretical and policy implications, states the review's limitations, and assesses how far the gap has been closed; the Conclusion summarizes the argument. 2. Literature Review University engagement and the organization of collaboration The theoretical starting point for most work on university partnerships is the Triple Helix model, which describes innovation as the product of interdependent relations among universities, industry, and government rather than of any single institutional sphere (Etzkowitz & Leydesdorff, 2000). Carayannis and Campbell (2009) extended this reasoning with a Quadruple Helix that adds a fourth strand, the media-based and culture-based public, and with the claim that multiple modes of knowledge production coexist and co-evolve rather than replacing one another. Both ideas matter for agriculture, where public legitimacy and cultural attitudes toward food and technology shape what innovations are acceptable, a point to which the review returns. Within this broad frame, a distinct literature examines how collaboration is actually organized. Perkmann et al. (2013) drew an influential distinction between academic engagement, meaning collaborative research, contract research, and consulting sustained through relationships, and commercialization, meaning patenting, licensing, and venture creation; their follow-up review of work published between 2011 and 2019 documents how strongly research on engagement has grown as a field in its own right (Perkmann et al., 2021). D'Este and Patel (2007) showed for the United Kingdom that academics interact with industry through a much wider variety of channels than patenting statistics capture. Ankrah and Al-Tabbaa (2015) systematized the organizational forms, motivations, and process dynamics of collaboration, and Mascarenhas et al. (2018) and Skute et al. (2019) mapped the field's growth and its main research streams. Collectively these reviews establish that partnership form is consequential: different forms carry different kinds of knowledge and impose different governance demands. A second stream examines why collaboration succeeds or fails. Bruneel et al. (2010) showed that barriers between academic and industrial partners are not fixed features of the relationship but can be attenuated, drawing attention to relational assets such as trust. Rajalo and Vadi (2017) reconceptualized collaboration at the level of individual partnerships, arguing that productive interaction depends on conditions internal to each dyad rather than on formal agreements alone. The review by de Wit-de Vries et al. (2019) synthesized the mechanisms of knowledge transfer in university–industry research partnerships, underscoring that transfer is an accomplishment of process rather than an automatic by-product of contact. At a more aggregate level, Galan-Muros and Davey (2019) assembled an ecosystem framework for university–business cooperation, and Heaton et al. (2019) applied dynamic capabilities theory to argue that a university's capacity to sense opportunities, seize them, and reconfigure its own structures shapes its contribution to an innovation ecosystem; they add that the flexible, entrepreneurial university management this requires has received less scholarly attention than the Triple Helix model it presupposes. Compagnucci and Spigarelli (2020) documented both the potential and the constraints of the third mission, including tensions with universities' core teaching and research tasks. Agricultural innovation systems and the digital turn Agricultural scholarship approaches innovation from the opposite direction: not from the university outward but from the farming system inward. The agricultural innovation systems perspective, codified for investment practice by the World Bank (2012), treats innovation as the emergent outcome of interactions among research organizations, education, extension and advisory services, firms, and farmers, and directs attention to the institutions that link them. Pigford et al. (2018) pushed this perspective toward an innovation ecosystems approach oriented to niche design in sustainability transitions, and Klerkx and Begemann (2020) reframed it in mission-oriented terms, asking what directional missions for food systems imply for who participates in innovation and how. Digitalization has given this literature a strong contemporary focus. Wolfert et al. (2017) reviewed big data applications in smart farming; Klerkx et al. (2019) organized the emerging social science on digital agriculture, smart farming, and Agriculture 4.0 and set out a future research agenda for the field; Fielke et al. (2020) reviewed how digitalization reshapes agricultural knowledge and advice networks; and Ingram and Maye (2020) examined what digitalization means for agricultural knowledge itself, including the position of formal expertise. Basso and Antle (2020) argued that digital tools should be used to design sustainable agricultural systems rather than merely to optimize existing ones, a formulation that presupposes sustained interaction between researchers and practitioners rather than one-way technology delivery. In lower-income settings, Aker (2011) reviewed evidence on information and communication technologies as channels for agricultural extension, Deichmann et al. (2016) proposed a framework in which digital technology contributes through inclusion, efficiency, and lower transaction costs while cautioning that it removes only some of the barriers smallholders face, and Fabregas et al. (2019) reviewed digital agricultural advice and identified the financing of such services as an unresolved problem. Two further strands complete the picture. A responsibility strand argues that Agriculture 4.0 requires broader and earlier societal engagement than conventional technology assessment provides: Rose and Chilvers (2018) called for widening responsible innovation in smart farming, Eastwood et al. (2019) traced how responsible research and innovation can move from fragmented practices to a comprehensive approach in technology development programmes, and Klerkx and Rose (2020) argued that transition pathways should preserve diversity rather than lock in a single technological trajectory. A bundling strand holds that technological innovations transform food systems only when combined with social and institutional innovations (Barrett et al., 2020; Herrero et al., 2020). The unresolved intersection Read together, the two literatures leave a specific question open. The collaboration literature explains partnership forms and mechanisms but abstracts from agriculture's institutional particularities: dispersed and heterogeneous adopters, a central role for advisory intermediaries, strong public-interest stakes, and, in much of the world, smallholder production under infrastructure constraints. The agricultural literature explains those particularities but does not differentiate among the university partnership forms operating within them. This is an interpretive assessment of the corpus reviewed here rather than a finding of any single study, and it defines the task of this review: to specify the partnership configurations, mechanisms, and conditions jointly, so that findings from each literature can discipline the other. 3. Method Review design and rationale The review follows the integrative method, which combines diverse designs and literatures in order to generate new frameworks rather than to estimate effects (Torraco, 2005; Whittemore & Knafl, 2005). An integrative design was chosen over a systematic-review design for two reasons. First, the relevant evidence is heterogeneous: it spans conceptual work in innovation studies, reviews and empirical studies in agricultural economics, and social science on digitalization, which no single effect-size logic could aggregate. Second, the review's purpose is theory construction, the purpose for which Snyder (2019) recommends integrative over systematic approaches. The design implements the stages that Whittemore and Knafl (2005) specify: problem identification, literature search, data evaluation, data analysis, and presentation. Search strategy Searches were conducted in Scopus and Web of Science, complemented by Google Scholar for coverage of institutional reports and forward and backward citation tracking. Search strings combined a university-partnership block with an agritech block. Representative strings were: ("university–industry" OR "academic engagement" OR "technology transfer" OR "university partnership" OR "third mission") AND (agritech OR "agricultural technology" OR "digital agriculture" OR "smart farming" OR "precision agriculture" OR "agricultural innovation"); and ("agricultural innovation system" OR "innovation ecosystem" OR "advisory services" OR extension) AND (university OR "higher education" OR research). The core publication window was 2016 to 2026, reflecting the period in which digital agriculture scholarship consolidated; seminal earlier works were retained where the conceptual argument required them, including foundational statements of the Triple Helix and integrative-review methodology. Searches were limited to English-language publications. Inclusion and exclusion criteria and screening Sources were included if they satisfied all of the following: (a) peer-reviewed journal article, or institutional publication from an international organization with documented review processes; (b) substantive treatment of at least one of the review's three constructs, namely university or research-organization partnerships, knowledge translation mechanisms, or agricultural innovation conditions; (c) conceptual, review, or empirical design with transparent method. Sources were excluded if they (a) addressed agricultural technology with no institutional or organizational content, such as purely agronomic or engineering studies; (b) addressed university collaboration only in teaching contexts without a research or innovation dimension; or (c) could not be verified against bibliographic records. Screening proceeded in two stages, title-and-abstract followed by full-record assessment, applying the criteria in the order listed. Because the search was iterative and citation tracking was used throughout, the review does not report a quantitative screening flow; the corpus is instead documented through full citation of every included source, each verified against its bibliographic record before inclusion. Analysis and framework derivation Data evaluation followed the integrative-review convention of appraising sources for relevance and methodological transparency rather than through a single quality-scoring instrument, since the corpus deliberately mixes conceptual, review, and empirical designs whose quality criteria differ (Whittemore & Knafl, 2005). Each source's bibliographic record was checked against its publisher or index entry, and the type of claim each source could support, conceptual proposition, single-study finding, or synthesized evidence, was recorded at the point of coding so that attribution strength in the synthesis would match the underlying evidence. Included sources were then coded into a concept matrix organized by the three research questions: the partnership forms a source described or theorized, the mechanisms it identified as carrying knowledge between partners, and the conditions it associated with success, failure, or exclusion. Constant comparison across the two parent literatures was then used to consolidate forms into configurations, to align mechanism vocabularies (for example, matching the collaboration literature's account of relational engagement with the agricultural literature's account of knowledge intermediation), and to distinguish moderating conditions from steering considerations. The typology in Table 1 is the direct output of the first coding dimension; the framework in Figure 1 integrates all three. Both are conceptual syntheses: they organize claims made in the cited literature and do not report new empirical estimates. Rigor and trustworthiness Three practices bounded bias. First, every reference was verified against its bibliographic record before citation, and claims are attributed at the strength the underlying source supports, with single-study findings reported as such. Second, the two parent literatures were searched and coded symmetrically, so that the framework is not an extension of one literature with the other as decoration. Third, the review's scope boundaries are explicit: it covers English-language scholarship, it prioritizes the 2016 to 2026 window, and its unit of analysis is the partnership configuration, not the individual technology or the individual scientist. Within these boundaries the protocol above is reproducible: another reviewer applying the same strings, window, and criteria would assemble a corpus with substantially the same composition, though the interpretive synthesis would remain a judgment for which the authorial reading is accountable. 4. Findings A typology of university partnership configurations in agritech (RQ1) The corpus yields five recurrent configurations through which universities engage in agritech innovation, summarized in Table 1. They differ in principal actors, in the mechanism that carries knowledge, and in the kind of outcome they are organized to produce. The first configuration is bilateral academic engagement: collaborative research, contract research, and consulting between university groups and agritech firms or producer organizations. This is the agricultural instance of the engagement channel that Perkmann et al. (2013, 2021) identify as the dominant mode of university–industry interaction generally, and the variety of its concrete channels mirrors the variety D'Este and Patel (2007) documented across sectors. The second configuration is commercialization: patenting, licensing, and spin-off formation around university-held technology, the channel that the same literature distinguishes sharply from engagement (Perkmann et al., 2013) and that third-mission scholarship shows to be demanding of specialized university structures (Compagnucci & Spigarelli, 2020). The third configuration is the mission-oriented consortium: multi-partner programmes in which government, universities, and industry align around a defined objective such as emissions reduction or food security, the form that mission-oriented agricultural innovation systems thinking anticipates (Klerkx & Begemann, 2020) and that inherits its logic from the Triple Helix (Etzkowitz & Leydesdorff, 2000). The fourth configuration is the ecosystem platform: a university acting as durable anchor of a regional innovation ecosystem that includes incubation, capability building, and orchestration of many partners, consistent with the ecosystem frameworks of Galan-Muros and Davey (2019) and the dynamic-capabilities account of Heaton et al. (2019), and with the niche-design role that Pigford et al. (2018) sketch for agricultural sustainability transitions. The fifth configuration is the digital extension partnership: universities working with advisory services, digital platform providers, and farmer organizations to deliver and adapt knowledge-intensive services, the form foregrounded by scholarship on advisory-network digitalization (Fielke et al., 2020; Ingram & Maye, 2020) and by the developing-country evidence on ICT-mediated extension (Aker, 2011; Fabregas et al., 2019). Three observations follow from the typology. First, the configurations are analytic rather than mutually exclusive: a single university may operate all five at once, and ecosystem platforms in particular often contain the other four as constituent activities. Second, the configurations are ordered by the breadth of the actor coalition they require, from the dyad of bilateral engagement to the multi-actor coalitions of platforms and extension partnerships; this ordering matters because the agricultural literature associates system-level outcomes with broader coalitions (Klerkx & Begemann, 2020; World Bank, 2012), while the collaboration literature shows that relational quality is easiest to sustain in narrow ones (Rajalo & Vadi, 2017). Partnership design in agritech therefore confronts a standing tension between relational depth and systemic reach. Third, the fifth configuration has no clear counterpart in the general university–industry literature; it is specific to sectors with strong advisory intermediation, which is one reason cross-sectoral reviews underdescribe agriculture. Its growing weight in the corpus reflects the digitalization of advice documented by Fielke et al. (2020), which repositions universities from suppliers of content for human advisers toward partners in the design and validation of digital advisory services themselves. Table 1 Typology of University Partnership Configurations in Agritech Innovation Configuration Principal actors Dominant translation mechanism Typical outcome orientation Illustrative sources Bilateral academic engagement University research group; agritech firm or producer organization Relational knowledge co-production Firm- or organization-level product and process innovation Perkmann et al. (2013); D'Este and Patel (2007); de Wit-de Vries et al. (2019) Commercialization University technology transfer function; ventures and investors Property-based transfer (patents, licences, spin-offs) New agritech products and ventures Perkmann et al. (2013); Compagnucci and Spigarelli (2020) Mission-oriented consortium Government, universities, industry, and research funders Agenda alignment and pooled research capacity Directional, system-level innovation objectives Klerkx and Begemann (2020); Etzkowitz and Leydesdorff (2000); World Bank (2012) Ecosystem platform University as anchor with regional firms, incubators, and intermediaries Capability building and ecosystem orchestration Regional agritech ecosystem development Heaton et al. (2019); Galan-Muros and Davey (2019); Pigford et al. (2018) Digital extension partnership University, advisory services, platform providers, and farmer organizations Co-design and digital knowledge intermediation Farm-level adoption and inclusive advisory reach Fielke et al. (2020); Ingram and Maye (2020); Aker (2011); Fabregas et al. (2019) Note. The typology is an analytic synthesis of the sources cited in each row; configurations co-occur in practice, and a single university may operate several simultaneously. Mechanism and outcome labels follow the framework developed in the Findings section. Mechanisms that translate research into innovation (RQ2) Across configurations, four translation mechanisms recur. The first is relational knowledge co-production. Reviews of knowledge transfer and academic engagement place sustained interpersonal relationships at the center of how knowledge moves between universities and firms (de Wit-de Vries et al., 2019; Perkmann et al., 2013); Bruneel et al. (2010) linked the attenuation of barriers between the academic and commercial spheres to relational factors such as trust; and Rajalo and Vadi (2017) located the conditions for productive interaction inside the individual partnership rather than in formal agreements. Agricultural evidence points the same way: advisory relationships remain central even as their medium digitalizes (Fielke et al., 2020; Ingram & Maye, 2020). The second mechanism is property-based transfer: the codification of research into intellectual property that firms or ventures can carry to market. It is the mechanism proper to the commercialization configuration, and the literature warns against equating it with university impact in general, since it captures a narrow slice of interaction (D'Este & Patel, 2007; Perkmann et al., 2013). The third mechanism is capability building. Ecosystem-level frameworks converge on the point that partnerships produce not only artifacts but capacities: absorptive capacity in firms, entrepreneurial capability in regions, and adaptive capability in universities themselves (Galan-Muros & Davey, 2019; Heaton et al., 2019). In agriculture this extends to farmer capability, which conditions whether knowledge-intensive technologies are usable at all; the contrast Lowenberg-DeBoer and Erickson (2019) draw between rapidly adopted automation, which embeds knowledge in the machine, and slowly adopted variable-rate management, which demands knowledge in the user, illustrates how outcomes depend on where capability resides. The fourth mechanism is co-design: the joint specification of technologies and services with their eventual users. The responsibility literature treats co-design as the practical core of responsible innovation in smart farming (Eastwood et al., 2019; Rose & Chilvers, 2018), and Fabregas et al. (2019) point to locally relevant advice as central to the promise of digital extension. Interpreted through the framework developed here, co-design is the mechanism by which the fifth configuration compensates for the distance between laboratory and field that the other mechanisms leave open. The alignment of mechanism with configuration is the review's central analytical claim: each configuration is, in effect, an institutional casing for one dominant mechanism, and partnership designs fail when the casing and the mechanism are mismatched, as when a co-design problem is handled through property-based transfer. Contextual conditions across settings (RQ3) The mechanisms do not operate uniformly. Four conditions recur in the corpus as moderators. The first is institutional distance within the partnership: differences in incentives, time horizons, and disclosure norms between academic and commercial partners, which the collaboration literature identifies as the root of persistent barriers (Bruneel et al., 2010; Rajalo & Vadi, 2017). The second is the strength of intermediary institutions. Where advisory systems are dense, universities can reach dispersed users through them; where digitalization reconfigures advisory networks, the position of formal research knowledge itself shifts (Fielke et al., 2020; Ingram & Maye, 2020). The third is infrastructure and user heterogeneity. Digital technologies presuppose connectivity, complementary inputs, and minimum scale; Deichmann et al. (2016) caution that they remove only some of the barriers facing poorer farmers, and Fabregas et al. (2019) identify the financing of digital advisory services as unresolved, both of which explain why pilot successes scale unevenly in low- and middle-income countries. The fourth is public legitimacy. The Quadruple Helix argument that innovation systems answer to a media- and culture-based public (Carayannis & Campbell, 2009) has particular force in food and agriculture, where societal acceptance conditions technology pathways, a point the responsible-innovation strand develops in detail (Klerkx & Rose, 2020; Rose & Chilvers, 2018). Data governance illustrates the interaction of these conditions: reviews of smart farming identify questions of data ownership and control among the issues that accompany big data applications (Wolfert et al., 2017), and such questions simultaneously raise institutional distance within partnerships, burden intermediaries who must broker data relationships, and put legitimacy at stake with farming communities. Conditions of this kind are therefore best read as a configuration of pressures rather than as a checklist of separate obstacles. The high-income and low- and middle-income contrast, on this reading, is not a difference in kind but a difference in which conditions bind. In high-income settings the binding constraints cluster around institutional distance and legitimacy; in low- and middle-income settings they cluster around intermediary strength, infrastructure, and financing, with the agricultural innovation systems tradition supplying the investment logic for relaxing them (World Bank, 2012). Directionality and responsibility as steering considerations A final theme cuts across the other three. Recent agricultural scholarship insists that the question is not only whether partnerships produce innovation but which innovation they produce. Mission-oriented framings ask partnerships to serve defined societal objectives (Klerkx & Begemann, 2020); transition scholarship warns against locking food systems into a single technological pathway (Klerkx & Rose, 2020); and the bundling argument holds that technological innovations change food systems only when combined with social and institutional innovations (Barrett et al., 2020; Herrero et al., 2020). For university partnerships, directionality is therefore not an external policy constraint but a design parameter: it determines which configurations are appropriate, since missions favour consortia and platforms, and which mechanisms must be present, since responsibility requirements presuppose co-design (Eastwood et al., 2019). Figure 1 integrates the four findings. Partnership configurations, on the left, translate academic research into agritech innovation outcomes, on the right, through the four mechanisms in the center. Contextual conditions moderate every configuration-mechanism-outcome pathway from below; directionality and responsibility steer the selection of configurations and mechanisms from above; and outcomes feed back into future partnership formation. The framework is conceptual: it organizes verified claims from the reviewed literature into a testable structure and makes no empirical estimate of pathway strength. Figure 1. Conceptual framework linking university partnership configurations to agritech innovation outcomes. The framework is a conceptual synthesis of the reviewed literature; arrows denote proposed pathways, not estimated effects. 5. Discussion Theoretical implications The review carries three implications for theory. First, it shows that the university–industry collaboration literature's central distinction, engagement versus commercialization (Perkmann et al., 2013), is necessary but insufficient for agriculture: a sector with strong advisory intermediation requires at least the fifth configuration, digital extension partnership, and arguably treats it as the primary channel to the majority of users. Extending collaboration theory with sector-specific configurations is a more promising route than treating sector as a control variable. Second, the framework specifies where the two parent literatures connect: mechanisms are the join. The collaboration literature theorizes mechanisms with partnership-level precision but weak context; the agricultural literature theorizes context with system-level precision but undifferentiated partnerships; aligning both around mechanisms lets each discipline the other. Third, the analysis supports reading responsibility and directionality as design parameters of partnership rather than as external evaluation criteria, which connects responsible-innovation scholarship (Eastwood et al., 2019; Rose & Chilvers, 2018) to the organizational literature it has so far run parallel to. Practical and policy implications For university leaders, the typology functions as a portfolio instrument. Because engagement, not commercialization, is the channel the collaboration literature treats as primary (Perkmann et al., 2013, 2021), and because the adoption evidence associates diffusion of knowledge-intensive technologies with user capability (Lowenberg-DeBoer & Erickson, 2019), the synthesis implies that agritech strategies built mainly around patenting and spin-offs address a narrow part of the problem; portfolios should be weighted toward engagement, platform, and extension configurations, with commercialization reserved for technologies that genuinely travel as property. For policymakers and funders, the framework identifies the moderating conditions as the targets of public investment: intermediary institutions, connectivity, and service financing in low- and middle-income settings (Deichmann et al., 2016; Fabregas et al., 2019; World Bank, 2012), and boundary-spanning arrangements that reduce institutional distance in high-income settings (Bruneel et al., 2010). For both audiences, the directionality theme implies that partnership instruments should specify missions and inclusion requirements at the design stage rather than auditing them after the fact (Klerkx & Begemann, 2020). A further implication concerns measurement. Because the configurations carry different mechanisms, they require different indicators of success: licensing income is informative for the commercialization configuration but nearly meaningless for extension partnerships, whose contribution registers in advisory reach and user capability. Evaluation regimes that apply commercialization metrics across a whole portfolio will systematically undervalue the configurations that the agricultural evidence identifies as reaching most users, a measurement bias that third-mission scholarship has flagged in general terms (Compagnucci & Spigarelli, 2020) and that acquires particular weight in agriculture. Limitations The review's limitations follow from its method. It is an integrative synthesis, not a systematic review: it reports no quantitative screening flow, and corpus assembly involved iterative citation tracking in which authorial judgment shaped inclusion at the margin. Its evidence base is English-language and journal-weighted, which underrepresents grey literature from national innovation agencies and non-anglophone scholarship, both of which are likely to matter in agriculture. The framework is derived from literature that is itself uneven: the collaboration literature draws disproportionately on high-income economies, so the low- and middle-income cells of the synthesis rest on a thinner and more advisory-focused evidence base. Finally, the framework's pathways are conceptual propositions; the review provides no empirical test of their relative strength, and single-study findings cited in support, such as the adoption contrast reported by Lowenberg-DeBoer and Erickson (2019), should not be read as consensus estimates. Future research Three lines of work follow directly from what remains open. First, comparative empirical studies should test the typology's central claim, that configuration-mechanism fit predicts partnership outcomes, across sectors and income settings; the concept matrix underlying Table 1 offers a coding frame for such designs. Second, the fifth configuration needs dedicated study: research on how universities participate in digitalized advisory networks, and on sustainable financing models for digital extension, would address the open problem identified by Fabregas et al. (2019). Third, the responsibility strand would benefit from organizational specificity: studies of how mission requirements and inclusion criteria are written into partnership agreements, and with what effect, would connect responsible-innovation theory to partnership practice (Eastwood et al., 2019; Klerkx & Rose, 2020). With respect to the gap stated in the Introduction, the review closes its conceptual portion: it specifies the configurations, mechanisms, and conditions that neither parent literature specified jointly, and it does so from verified published evidence. The empirical portion of the gap, measurement of pathway strength and comparative performance, remains open and defines the agenda above. 6. Conclusion This article asked how university partnerships advance agritech innovation. An integrative review of the university–industry collaboration and agricultural innovation literatures produced a typology of five partnership configurations, bilateral engagement, commercialization, mission-oriented consortia, ecosystem platforms, and digital extension partnerships, and a conceptual framework in which these configurations translate academic research into innovation outcomes through four mechanisms: relational co-production, property-based transfer, capability building, and co-design. Contextual conditions, including institutional distance, intermediary strength, infrastructure, and public legitimacy, moderate the pathways, and directionality and responsibility considerations steer configuration choice. The central conclusion is that partnership form must match the knowledge mechanism it is meant to carry, and that in agriculture the match depends on conditions, especially advisory intermediation and user capability, that general collaboration theory does not model. Universities that treat their agritech engagement as a portfolio across configurations, and funders that invest in the conditions rather than only in the technologies, act in line with the synthesized evidence. The framework offered here is a structure for the comparative research that should now test it. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics. This study is a review of published literature and did not involve human participants, animal subjects, or personal data; ethical approval was therefore not required. Data Availability. No new data were created or analyzed in this study. All sources synthesized are cited in the reference list and publicly available through the digital object identifiers provided. References Aker, J. C. (2011). Dial "A" for agriculture: A review of information and communication technologies for agricultural extension in developing countries. Agricultural Economics, 42(6), 631–647. https://doi.org/10.1111/j.1574-0862.2011.00545.x Ankrah, S., & Al-Tabbaa, O. (2015). Universities–industry collaboration: A systematic review. Scandinavian Journal of Management, 31(3), 387–408. https://doi.org/10.1016/j.scaman.2015.02.003 Barrett, C. B., Benton, T. G., Cooper, K. A., Fanzo, J., Gandhi, R., Herrero, M., James, S., Kahn, M., Mason-D'Croz, D., Mathys, A., Nelson, R. J., Shen, J., Thornton, P., Bageant, E., Fan, S., Mude, A. G., Sibanda, L. M., & Wood, S. (2020). Bundling innovations to transform agri-food systems. Nature Sustainability, 3(12), 974–976. https://doi.org/10.1038/s41893-020-00661-8 Basso, B., & Antle, J. (2020). Digital agriculture to design sustainable agricultural systems. Nature Sustainability, 3(4), 254–256. https://doi.org/10.1038/s41893-020-0510-0 Bruneel, J., D'Este, P., & Salter, A. (2010). Investigating the factors that diminish the barriers to university–industry collaboration. Research Policy, 39(7), 858–868. https://doi.org/10.1016/j.respol.2010.03.006 Carayannis, E. G., & Campbell, D. F. J. (2009). 'Mode 3' and 'Quadruple Helix': Toward a 21st century fractal innovation ecosystem. International Journal of Technology Management, 46(3/4), 201–234. https://doi.org/10.1504/IJTM.2009.023374 Compagnucci, L., & Spigarelli, F. (2020). The Third Mission of the university: A systematic literature review on potentials and constraints. Technological Forecasting and Social Change, 161, Article 120284. https://doi.org/10.1016/j.techfore.2020.120284 Deichmann, U., Goyal, A., & Mishra, D. (2016). Will digital technologies transform agriculture in developing countries? Agricultural Economics, 47(S1), 21–33. https://doi.org/10.1111/agec.12300 D'Este, P., & Patel, P. (2007). University–industry linkages in the UK: What are the factors underlying the variety of interactions with industry? Research Policy, 36(9), 1295–1313. https://doi.org/10.1016/j.respol.2007.05.002 de Wit-de Vries, E., Dolfsma, W. A., van der Windt, H. J., & Gerkema, M. P. (2019). Knowledge transfer in university–industry research partnerships: A review. The Journal of Technology Transfer, 44(4), 1236–1255. https://doi.org/10.1007/s10961-018-9660-x Eastwood, C., Klerkx, L., Ayre, M., & Dela Rue, B. (2019). Managing socio-ethical challenges in the development of smart farming: From a fragmented to a comprehensive approach for responsible research and innovation. Journal of Agricultural and Environmental Ethics, 32(5–6), 741–768. https://doi.org/10.1007/s10806-017-9704-5 Etzkowitz, H., & Leydesdorff, L. (2000). The dynamics of innovation: From National Systems and "Mode 2" to a Triple Helix of university–industry–government relations. Research Policy, 29(2), 109–123. https://doi.org/10.1016/S0048-7333(99)00055-4 Fabregas, R., Kremer, M., & Schilbach, F. (2019). Realizing the potential of digital development: The case of agricultural advice. Science, 366(6471), Article eaay3038. https://doi.org/10.1126/science.aay3038 Fielke, S., Taylor, B., & Jakku, E. (2020). Digitalisation of agricultural knowledge and advice networks: A state-of-the-art review. Agricultural Systems, 180, Article 102763. https://doi.org/10.1016/j.agsy.2019.102763 Galan-Muros, V., & Davey, T. (2019). The UBC ecosystem: Putting together a comprehensive framework for university–business cooperation. The Journal of Technology Transfer, 44(4), 1311–1346. https://doi.org/10.1007/s10961-017-9562-3 Heaton, S., Siegel, D. S., & Teece, D. J. (2019). Universities and innovation ecosystems: A dynamic capabilities perspective. Industrial and Corporate Change, 28(4), 921–939. https://doi.org/10.1093/icc/dtz038 Herrero, M., Thornton, P. K., Mason-D'Croz, D., Palmer, J., Benton, T. G., Bodirsky, B. L., Bogard, J. R., Hall, A., Lee, B., Nyborg, K., Pradhan, P., Bonnett, G. D., Bryan, B. A., Campbell, B. M., … West, P. C. (2020). Innovation can accelerate the transition towards a sustainable food system. Nature Food, 1(5), 266–272. https://doi.org/10.1038/s43016-020-0074-1 Ingram, J., & Maye, D. (2020). What are the implications of digitalisation for agricultural knowledge? Frontiers in Sustainable Food Systems, 4, Article 66. https://doi.org/10.3389/fsufs.2020.00066 Klerkx, L., & Begemann, S. (2020). Supporting food systems transformation: The what, why, who, where and how of mission-oriented agricultural innovation systems. Agricultural Systems, 184, Article 102901. https://doi.org/10.1016/j.agsy.2020.102901 Klerkx, L., Jakku, E., & Labarthe, P. (2019). A review of social science on digital agriculture, smart farming and agriculture 4.0: New contributions and a future research agenda. NJAS: Wageningen Journal of Life Sciences, 90–91, Article 100315. https://doi.org/10.1016/j.njas.2019.100315 Klerkx, L., & Rose, D. (2020). Dealing with the game-changing technologies of Agriculture 4.0: How do we manage diversity and responsibility in food system transition pathways? Global Food Security, 24, Article 100347. https://doi.org/10.1016/j.gfs.2019.100347 Lowenberg-DeBoer, J., & Erickson, B. (2019). Setting the record straight on precision agriculture adoption. Agronomy Journal, 111(4), 1552–1569. https://doi.org/10.2134/agronj2018.12.0779 Mascarenhas, C., Ferreira, J. J., & Marques, C. (2018). University–industry cooperation: A systematic literature review and research agenda. Science and Public Policy, 45(5), 708–718. https://doi.org/10.1093/scipol/scy003 Perkmann, M., Salandra, R., Tartari, V., McKelvey, M., & Hughes, A. (2021). Academic engagement: A review of the literature 2011–2019. Research Policy, 50(1), Article 104114. https://doi.org/10.1016/j.respol.2020.104114 Perkmann, M., Tartari, V., McKelvey, M., Autio, E., Broström, A., D'Este, P., Fini, R., Geuna, A., Grimaldi, R., Hughes, A., Krabel, S., Kitson, M., Llerena, P., Lissoni, F., Salter, A., & Sobrero, M. (2013). Academic engagement and commercialisation: A review of the literature on university–industry relations. Research Policy, 42(2), 423–442. https://doi.org/10.1016/j.respol.2012.09.007 Pigford, A.-A. E., Hickey, G. M., & Klerkx, L. (2018). Beyond agricultural innovation systems? Exploring an agricultural innovation ecosystems approach for niche design and development in sustainability transitions. Agricultural Systems, 164, 116–121. https://doi.org/10.1016/j.agsy.2018.04.007 Rajalo, S., & Vadi, M. (2017). University-industry innovation collaboration: Reconceptualization. Technovation, 62–63, 42–54. https://doi.org/10.1016/j.technovation.2017.04.003 Rose, D. C., & Chilvers, J. (2018). Agriculture 4.0: Broadening responsible innovation in an era of smart farming. Frontiers in Sustainable Food Systems, 2, Article 87. https://doi.org/10.3389/fsufs.2018.00087 Skute, I., Zalewska-Kurek, K., Hatak, I., & de Weerd-Nederhof, P. (2019). Mapping the field: A bibliometric analysis of the literature on university–industry collaborations. The Journal of Technology Transfer, 44(3), 916–947. https://doi.org/10.1007/s10961-017-9637-1 Snyder, H. (2019). Literature review as a research methodology: An overview and guidelines. Journal of Business Research, 104, 333–339. https://doi.org/10.1016/j.jbusres.2019.07.039 Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367. https://doi.org/10.1177/1534484305278283 Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Wolfert, S., Ge, L., Verdouw, C., & Bogaardt, M.-J. (2017). Big data in smart farming: A review. Agricultural Systems, 153, 69–80. https://doi.org/10.1016/j.agsy.2017.01.023 World Bank. (2012). Agricultural innovation systems: An investment sourcebook. World Bank. https://doi.org/10.1596/978-0-8213-8684-2 Hashtags: #SDG2 #ZeroHunger #SDG9 #IndustryInnovationAndInfrastructure #SDG17 #PartnershipsForTheGoals #SDG4 #AgritechInnovation #UniversityIndustryCollaboration #TechnologyTransfer #DigitalAgriculture #AgriculturalInnovationSystems #IntegrativeReview #U7YJournal #U7Y #AcademicResearch

Author: Lucas Silva Affiliation: Swiss International University (SIU) ORCID ID: 0009-0001-8067-5926 Submitted 07 April 2026; Revised 09 June 2026; Accepted 16 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10002 Volume 3, December 2026, (SpecSDG10002) Abstract Tropical deforestation persists despite two decades of rapid progress in satellite monitoring, and regulators increasingly depend on algorithmic mapping systems they did not build and cannot fully audit. This article presents an integrative review of the peer-reviewed literature on artificial intelligence (AI) and spatial computing for deforestation mapping, and of the governance research that examines how such systems inform regulatory engagement. Drawing on verified studies spanning remote-sensing science, geospatial platform engineering, and environmental governance, the review pursues three questions: which AI methods and spatial-computing infrastructures underpin operational deforestation mapping; through which mechanisms mapped information reaches regulatory processes; and under which conditions algorithmic monitoring changes regulatory outcomes rather than merely documenting loss. The synthesis shows that deep learning and cloud-based geospatial platforms have made near-real-time, high-resolution forest disturbance detection operational at continental scale, while the regulatory effect of these capabilities is conditional on institutional capacity, legal definitions that rarely match what satellites measure, and the distribution of compliance burdens. The article contributes a five-stage conceptual framework, from observation infrastructure through AI inference and platform dissemination to regulatory engagement mechanisms and accountability conditions, that connects detection research to governance research. The framework clarifies why technically accurate systems can fail as regulatory instruments and identifies design choices that improve the fit between alerts and enforcement, due diligence, and community monitoring. Implications are drawn for regulators implementing due-diligence legislation, for national monitoring programmes, and for the research agenda on accountable algorithmic environmental monitoring. Keywords: deforestation monitoring, deep learning, remote sensing, near-real-time alerts, spatial computing, environmental governance, supply-chain regulation 1. Introduction Agricultural expansion remains the dominant proximate cause of tropical forest loss. Pendrill et al. (2022) estimated that between 90% and 99% of deforestation across the tropics in 2011 to 2015 was driven by agriculture, although only 45% to 65% of the cleared land entered productive agricultural use within a few years. Curtis et al. (2018) found that roughly one quarter of global forest loss since 2001 reflected commodity-driven conversion to permanent land uses, and that the rate of such conversion had not declined over their study period. Losses are concentrated where forests matter most for carbon and biodiversity: Turubanova et al. (2018) documented continuing primary forest loss in Brazil, the Democratic Republic of the Congo, and Indonesia, and Lapola et al. (2023) estimated that around 38% of the remaining Amazon forest is already degraded by fire, edge effects, timber extraction, or extreme drought, with degradation emissions of up to 0.2 petagrams of carbon per year. The Global Forest Resources Assessment confirms that forest area continues to contract globally even as monitoring capability improves (FAO, 2020). The monitoring side of this problem has changed beyond recognition. Hansen et al. (2013) produced the first high-resolution global maps of twenty-first-century forest cover change from the Landsat archive, and the free availability of that archive has since anchored an expanding programme of forest change science (Wulder et al., 2019). Algorithmic advances moved the field from periodic mapping to continuous observation (Verbesselt et al., 2012; Zhu & Woodcock, 2014), deep learning improved the extraction of change signals from dense image streams (Ma et al., 2019; Zhu et al., 2017), and cloud-based geospatial platforms made planetary-scale analysis available far beyond specialist laboratories (Gorelick et al., 2017). Operational alert systems now report forest disturbance within days across the humid tropics (Hansen et al., 2016; Reiche et al., 2021), and national programmes such as Brazil's DETER system route alerts directly to enforcement agencies (Assunção et al., 2023; Diniz et al., 2015). Regulation has begun to presume this capability. The European Union Deforestation Regulation (EUDR) requires companies placing certain commodities on the EU market to demonstrate, with geolocation evidence, that products are deforestation-free, a shift that Berning and Sotirov (2023) characterize as a hardening of corporate accountability from voluntary market instruments toward state-based regulation. Brazil's enforcement economy has been shown to depend measurably on satellite alerts (Assunção et al., 2023), and supply-chain transparency initiatives increasingly treat mapped forest change as the evidentiary basis for excluding non-compliant producers (Gardner et al., 2019; Rajão et al., 2020). The research gap this article addresses sits between two mature but weakly connected literatures. Remote-sensing research on AI-based deforestation detection is extensive and methodologically sophisticated, yet it evaluates systems almost exclusively by classification accuracy and latency (de Bem et al., 2020; Masolele et al., 2021; Reiche et al., 2021). Governance research on deforestation regulation is equally developed, yet it typically treats monitoring data as an exogenous input rather than as the product of contestable algorithmic choices (Bager et al., 2021; Berning & Sotirov, 2023; Zhunusova et al., 2022). The small body of causal evidence that bridges the two shows that the link between detection and outcome is conditional rather than automatic: alert subscriptions reduced deforestation probability in Africa by 18% but showed no comparable effect on other continents (Moffette et al., 2021), and a randomized experiment in Peru found only imprecisely estimated reductions from community alert use (Slough et al., 2021). What the literature has not resolved is how the technical properties of AI mapping systems, the architecture of the platforms that distribute their outputs, and the institutional mechanisms of regulation jointly determine whether detection becomes engagement. This matters for regulators now implementing due-diligence law on the assumption that maps settle compliance questions, for monitoring programmes deciding where to invest scarce development effort, and for producer-country communities that bear the consequences of algorithmic error. This article responds with an integrative review and a conceptual contribution. It synthesizes verified peer-reviewed evidence across the detection, platform, and governance literatures and derives a five-stage framework that traces the path from observation infrastructure through AI inference and spatial-computing dissemination to regulatory engagement mechanisms and the accountability conditions that govern them. Three research questions structure the review. RQ1: Which AI methods and spatial-computing infrastructures underpin current operational deforestation mapping and alerting? RQ2: Through which mechanisms do mapped deforestation data inform regulatory engagement? RQ3: Under which conditions does algorithmic monitoring change regulatory outcomes, and which accountability problems constrain it? The remainder of the article reviews the literature thematically, describes the review method, presents findings organized around the three questions, and discusses theoretical and policy implications, limitations, and future research. 2. Literature Review From Periodic Mapping to Continuous Observation The empirical basis for all deforestation regulation is the measurement of forest change, and the measurement regime has shifted from occasional national inventories toward continuous satellite observation. Hansen et al. (2013) demonstrated that global, spatially explicit forest change mapping was feasible at Landsat resolution, a result made possible by the open Landsat archive whose scientific status Wulder et al. (2019) review in detail. Methodologically, the decisive move was temporal: instead of comparing two images, algorithms began to model entire time series. Verbesselt et al. (2012) showed that structural change detection in satellite time series could support near-real-time disturbance flagging, and Zhu and Woodcock (2014) generalized this logic into continuous change detection and classification using all available Landsat observations. Applied to the tropics, these advances revealed both ongoing primary forest loss in the largest forest nations (Turubanova et al., 2018) and a degradation problem of comparable magnitude to outright clearing: Bullock et al. (2020) estimated that degradation and natural disturbance affected roughly as much Amazonian forest area between 1995 and 2017 as deforestation, with about 17% of the original forest area disturbed by 2017. Institutional assessments such as FAO (2020) continue to provide the definitional and statistical baseline against which satellite products are interpreted. Machine Learning and Deep Learning for Forest Change Detection A second research stream concerns the inference methods applied to observation streams. Reviews by Zhu et al. (2017) and Ma et al. (2019) document the migration of deep learning from computer vision into remote sensing, covering classification, object detection, and semantic segmentation. In deforestation applications specifically, convolutional architectures have outperformed classical machine learning: de Bem et al. (2020) reported that encoder-decoder networks such as ResUnet exceeded random forest and multilayer perceptron baselines for Amazon deforestation mapping from Landsat, reaching values near 0.94 on kappa, F1, and intersection-over-union metrics. Machine learning has also extended detection into signals that visual interpretation misses, including tropical selective logging (Hethcoat et al., 2019), and into post-clearing questions, with Masolele et al. (2021) using spatial and temporal deep learning on Landsat time series to classify the land uses that follow deforestation across the tropics. The frontier has moved from detection toward anticipation: Ball et al. (2022) trained convolutional networks on two decades of Peruvian Amazon data to forecast the following year's deforestation at 30-metre resolution, achieving an F1 score of 0.71 and showing that networks can identify risk precursors such as newly built access routes. The stream's shared characteristic is evaluative: performance is defined by accuracy against reference data, not by consequences for governance. Spatial Computing as Infrastructure Detection at scale requires more than algorithms. Shekhar et al. (2015) define spatial computing as the set of ideas and technologies that connect computation to physical space, spanning positioning, geographic information systems, and spatial data management. In environmental monitoring, the practical embodiment of spatial computing is the cloud geospatial platform. Gorelick et al. (2017) describe Google Earth Engine as a planetary-scale analysis system co-locating the satellite archive with parallel computation, and Tamiminia et al. (2020) show in a systematic review that land cover and change detection dominate its scientific use. Gomes et al. (2020) compare seven platforms for big Earth-observation data, including Earth Engine, Sentinel Hub, Open Data Cube, SEPAL, openEO, JEODPP, and pipsCloud, highlighting differences in openness, data models, and processing paradigms. These infrastructures matter for regulation because they determine who can compute what: alert systems such as those examined below are joint products of sensor constellations, archives, and platform capacity rather than of algorithms alone. Forest Governance and Regulatory Instruments The governance literature approaches deforestation from the side of institutions. Brazil's experience is the central empirical case: Nepstad et al. (2014) attribute the roughly 70% decline in Brazilian Amazon deforestation after 2004 to a combination of enforcement, credit restrictions, protected areas, and supply-chain interventions in beef and soy, and Assunção et al. (2015) estimate that the policies of the 2000s avoided around 73,000 square kilometres of clearing. Supply-chain governance research examines how transparency initiatives reshape accountability among traders, financiers, and producers (Gardner et al., 2019), and how illegality concentrates among a subset of properties whose output nonetheless contaminates aggregate exports (Rajão et al., 2020). At the demand end, Bager et al. (2021) catalogue 86 policy options available to the EU for reducing imported deforestation, of which the EUDR's due-diligence obligation is the one enacted; Berning and Sotirov (2023) analyse that regulation as a deliberate hardening of accountability after perceived failures of voluntary instruments, while Zhunusova et al. (2022) warn that its compliance architecture may disadvantage smallholders, indigenous peoples, and local communities in producer countries. This literature is attentive to power and institutions but generally silent on the algorithmic provenance of the forest data on which the instruments rely. Read together, the streams converge on a specific unresolved question. Detection research demonstrates capability, platform research demonstrates scale, and governance research demonstrates institutional effects, but the conditional pathway from algorithmic output to regulatory consequence has been examined only in fragments (Assunção et al., 2023; Moffette et al., 2021; Slough et al., 2021; Tabor & Holland, 2021). The review below is designed to assemble those fragments into an explicit framework. 3. Method Design and Rationale The study is an integrative literature review with a conceptual contribution. The integrative format was chosen because the evidence base is heterogeneous across disciplines, spanning algorithm evaluations, platform descriptions, econometric policy studies, and legal-institutional analyses, and because the aim is theory building rather than effect-size estimation. Integrative reviews are the appropriate design when diverse methodologies and both empirical and theoretical sources must be combined into a new conceptualization (Whittemore & Knafl, 2005), and the review followed the staged logic of problem identification, literature search, evaluation, analysis, and synthesis recommended for such work (Torraco, 2005). A systematic review with meta-analysis was rejected because the outcome metrics of the constituent literatures are incommensurable; a purely conceptual essay was rejected because the framework needed grounding in verified empirical findings. The three research questions stated in the Introduction structure the protocol: RQ1 is answered through the detection and platform literature clusters, RQ2 through the governance cluster and the cross-cluster synthesis, and RQ3 through the subset of studies that evaluate monitoring against institutional outcomes. Search Strategy Literature was identified through structured searches of Scopus, Web of Science, and Google Scholar, complemented by backward and forward citation chasing from anchor papers. Example search strings combined method terms and domain terms: ("deep learning" OR "machine learning" OR "convolutional neural network") AND (deforestation OR "forest change" OR "forest disturbance") AND (Landsat OR Sentinel); ("near-real-time" OR alert*) AND (deforestation OR "forest loss") AND (monitoring OR enforcement); and (deforestation) AND ("due diligence" OR "supply chain" OR regulation OR governance) AND (satellite OR "remote sensing" OR transparency). The core search window was 2012 to 2026, reflecting the period in which time-series detection, deep learning, and cloud platforms became operational; seminal earlier methodological works were admitted where the argument required them. Searches were run iteratively between the thematic analysis stages so that categories emerging from one stream could be checked against the others. Inclusion and Exclusion Criteria Sources were included if they (a) were peer-reviewed journal articles, or institutional assessments from mandated international organizations; (b) addressed at least one of the three review constructs, namely AI or algorithmic detection of forest change, spatial-computing platforms or alert systems, or regulatory and governance responses to deforestation; and (c) reported verifiable bibliographic metadata resolvable through a digital object identifier or an institutional record. Sources were excluded if they (a) were preprints, conference abstracts, or grey literature without institutional standing; (b) concerned forests only incidentally, for example generic land-cover classification without a forest change component; or (c) could not be verified against the publisher's record. Every reference cited in this article was individually verified against its Crossref registration record, and where available the publisher abstract, before inclusion; sources whose metadata could not be confirmed were discarded rather than corrected. Analysis and Framework Derivation Included sources were coded thematically in three passes. The first pass assigned each source to one or more of the three construct clusters (detection, platform, governance). The second pass extracted, for each source, the claims relevant to the review questions: reported capabilities and limits for detection studies; coverage, cadence, and dissemination channels for platform studies; and mechanisms, conditions, and distributional effects for governance studies. The third pass compared claims across clusters to identify the points at which the literatures connect or fail to connect, following the constant-comparison logic of integrative synthesis (Torraco, 2005; Whittemore & Knafl, 2005). The five-stage framework presented in the Findings was derived abductively: stages were proposed where the cross-cluster comparison showed a distinct transformation of information (from radiance to classification, from classification to alert, from alert to institutional act), and each proposed stage was retained only if at least two independent verified sources evidenced both its operation and a failure mode specific to it. Rigor and Trustworthiness Several measures limited bias. All bibliographic records were verified against the Crossref registry rather than taken from secondary citations, which removes citation-drift error. Quantitative claims were carried into the synthesis only when the figure was observed in the verified record or abstract of the cited source; where a figure could not be confirmed, the claim was retained qualitatively or dropped. Claims are attributed at the strength of their evidence, distinguishing single-study findings from replicated patterns, and interpretive steps in the framework derivation are marked as conceptual rather than empirical. The scope is bounded in three ways that readers should weigh: the review privileges English-language journal literature; it emphasizes the tropical forest monitoring systems with the strongest documentation, which concentrates evidence on Brazil, the Congo Basin, and pan-tropical platforms; and it synthesizes published evaluations rather than conducting new accuracy or policy analysis. 4. Findings AI Methods for Deforestation Detection The first component of RQ1 concerns inference methods. Three generations of method are visible in the verified literature, and they coexist rather than replace one another. The first is dense time-series statistics on optical archives: structural break detection for near-real-time flagging (Verbesselt et al., 2012) and continuous change detection and classification across all available Landsat observations (Zhu & Woodcock, 2014). These methods established the temporal logic that later systems inherited, namely that disturbance is an anomaly in a modelled trajectory rather than a difference between two scenes. The second generation is supervised machine learning tuned to specific forest change signals, exemplified by the detection of tropical selective logging, a low-magnitude signal that conventional mapping misses (Hethcoat et al., 2019). The third is deep learning. Comparative evidence indicates a real but bounded advantage: in Amazon-wide Landsat experiments, convolutional segmentation networks outperformed random forest and perceptron baselines, with the strongest architecture reaching approximately 0.94 across kappa, F1, and intersection-over-union metrics (de Bem et al., 2020), consistent with the broader migration of deep learning through remote sensing documented in review work (Ma et al., 2019; Zhu et al., 2017). Two extensions of the deep learning generation matter specifically for regulation. Attribution methods classify what replaces forest: Masolele et al. (2021) demonstrated pan-tropical classification of post-deforestation land use from Landsat time series, information that due-diligence regimes need because their obligations attach to commodities rather than to clearing as such. Anticipation methods forecast where clearing will occur: Ball et al. (2022) showed that convolutional networks trained on two decades of Peruvian data could forecast the next year's deforestation at 30-metre resolution with an F1 score of 0.71, learning risk precursors such as new access routes directly from imagery. Against these gains stand persistent blind spots. Degradation, which affects Amazonian forest areas comparable to those deforested (Bullock et al., 2020) and now touches an estimated 38% of the remaining Amazon forest (Lapola et al., 2023), remains harder to detect than clearing because its spectral signal is subtle and transient. The answer to the methods half of RQ1 is therefore that operational capability is strongest exactly where regulatory definitions are simplest, on abrupt canopy removal, and weakest on the gradual processes that increasingly dominate forest carbon loss. Spatial-Computing Platforms and Alert Systems The infrastructure component of RQ1 concerns where these methods run and how their outputs travel. Cloud geospatial platforms co-locate archives and computation, which converted global-scale mapping from a data-logistics problem into an algorithm problem (Gorelick et al., 2017), and change detection is among the dominant scientific uses of such platforms (Tamiminia et al., 2020). Platform diversity is itself consequential: comparative analysis of seven major systems shows meaningful differences in openness, data models, and reproducibility (Gomes et al., 2020), which shape which states, companies, and communities can independently verify a map rather than merely consume it. Operational alert systems built on this infrastructure differ systematically in sensor basis, coverage, cadence, and institutional coupling, and these differences condition their regulatory usefulness. Table 1 synthesizes the principal documented systems from the verified literature. Landsat-based GLAD alerts extended disturbance alerting across the humid tropics (Hansen et al., 2016) and are distributed through the Global Forest Watch platform to subscribers in tropical countries (Moffette et al., 2021). Radar-based RADD alerts, introduced for the Congo Basin, exploit Sentinel-1 observations every 6 to 12 days at 10-metre resolution regardless of cloud cover, detected roughly four million disturbance events covering about 1.4 million hectares between January 2019 and July 2020, and resolve small-scale change, with around 80% of detected events smaller than half a hectare (Reiche et al., 2021). Brazil's DETER lineage is distinct in being state-operated and enforcement-coupled: DETER-B was designed by the national space institute INPE explicitly as a near-real-time detection system for the Brazilian Amazon (Diniz et al., 2015), operating alongside the annual official mapping that supports Brazil's deforestation statistics, and its alerts function as the operational trigger for federal enforcement (Assunção et al., 2023). The sensor bases of these systems are complementary rather than competing. Optical alerting inherits the archival depth and radiometric consistency of the Landsat programme (Wulder et al., 2019) but is interrupted by cloud, which matters doubly in the humid tropics: clouds delay detection, and, as the Brazilian evidence shows, cloud-induced gaps in alert coverage translate directly into gaps in enforcement presence (Assunção et al., 2023). Radar alerting removes the cloud constraint at the price of a shorter archive and different disturbance signatures (Reiche et al., 2021). Cadence interacts with the statistical logic of detection established in the time-series literature: declaring a disturbance from few observations trades speed against false alarms, so operational systems must choose a position on the timeliness-reliability frontier rather than optimize both (Verbesselt et al., 2012). These trade-offs are design decisions with regulatory consequences, because an enforcement agency acting on fast, low-confirmation alerts will visit more false positives, while a due-diligence auditor relying on slow, high-confirmation products may miss the clearing season entirely. The answer to the infrastructure half of RQ1 is that spatial computing has made latency, resolution, and coverage into design variables; what remains scarce is not detection capacity but institutional receptors for it. Table 1 Operational Satellite-Based Deforestation Monitoring Systems and Their Documented Regulatory Uses System Data basis Coverage Cadence Documented regulatory or governance use Sources Global forest change maps Landsat optical archive Global Annual Public transparency baseline underpinning platform analyses and supply-chain screening Hansen et al. (2013); Finer et al. (2018) GLAD alerts Landsat optical Humid tropics As new cloud-free observations arrive Alert subscriptions used by agencies and organizations across 22 tropical countries Hansen et al. (2016); Moffette et al. (2021) RADD alerts Sentinel-1 radar Congo Basin (initial deployment) Every 6 to 12 days at 10 m, cloud-independent Law-enforcement and forest-management support via Global Forest Watch Reiche et al. (2021) DETER and DETER-B Optical satellite imagery processed for rapid alerts Brazilian Amazon Near real time Directs federal environmental enforcement operations Diniz et al. (2015); Assunção et al. (2023) Community alert programmes Satellite-based tree cover loss alerts delivered to trained monitors Peruvian Amazon communities (documented case) Incentivized patrols by community monitors Community forest monitoring and territorial management Slough et al. (2021); Tabor & Holland (2021) Note. Compiled from the cited sources; coverage, cadence, and resolution figures are as reported therein. Systems are ordered by increasing coupling to a specific institutional user. From Maps to Regulatory Engagement RQ2 asks through which mechanisms mapped deforestation informs regulation. The verified literature supports a typology of four mechanisms, which the conceptual framework in Figure 1 places within a five-stage pipeline running from observation infrastructure through AI inference and platform dissemination to engagement and accountability. The first mechanism is enforcement targeting by states. The strongest causal evidence comes from Brazil, where Assunção et al. (2023) exploit cloud cover as an exogenous constraint on DETER's detection capacity and show that satellite-directed monitoring and enforcement effectively curb deforestation. This mechanism operates within a broader policy mix whose historical effect is well documented: the post-2004 combination of enforcement, credit restriction, protected areas, and supply-chain agreements accompanied a decline in Brazilian Amazon deforestation of about 70% (Nepstad et al., 2014), with the 2000s policy package estimated to have avoided roughly 73,000 square kilometres of clearing (Assunção et al., 2015). The second mechanism is transparency-based engagement through public platforms. Finer et al. (2018) describe the operational chain from satellite detection to field intervention, in which published alerts enable prosecutors, journalists, and civil-society organizations to compel responses that agencies might not initiate. Supply-chain transparency extends this mechanism to markets: property-level mapping allows traders and financiers to be confronted with the deforestation embedded in their sourcing (Gardner et al., 2019), and spatially explicit analysis shows that illegality is concentrated among a minority of properties whose output nonetheless taints aggregate exports, making property-resolved maps directly actionable for sectoral agreements (Rajão et al., 2020). The third mechanism is legal due diligence. The EUDR converts maps from advocacy material into compliance evidence: operators must geolocate production plots and demonstrate the absence of post-cutoff deforestation, an architecture that Berning and Sotirov (2023) interpret as hard, state-based accountability replacing voluntary certification. Among the wide option space available to importing jurisdictions (Bager et al., 2021), this design makes satellite mapping a de facto regulatory infrastructure, because both operators and competent authorities must interrogate forest change data to discharge statutory duties. Figure 1 depicts how these mechanisms sit within the full pipeline. Observation infrastructure (stage 1) comprises satellite constellations and open archives; AI inference (stage 2) transforms observations into classified change, attributed drivers, and forecast risk; platform dissemination (stage 3) turns classifications into subscribable alerts and auditable map services; regulatory engagement (stage 4) is where the four mechanisms operate; and accountability conditions (stage 5) feed back into every earlier stage, since legal admissibility, contestability, and equity requirements discipline what detection and dissemination must provide. The framework is conceptual, derived from the synthesis rather than tested against new data, but each stage and each feedback is anchored in the verified studies discussed in this section. The fourth mechanism is community-based monitoring. Alerts delivered to trained and incentivized community monitors in the Peruvian Amazon produced reductions in tree cover loss that were imprecisely estimated but institutionally meaningful, as monitors came to be regarded as forest management authorities within their communities (Slough et al., 2021). Design analysis of early warning systems similarly distinguishes rapid-response uses from targeted-response uses and stresses stakeholder engagement as the binding constraint (Tabor & Holland, 2021). Across all four mechanisms, the common structure is that the map does no regulatory work until a specific institution with standing, capacity, and incentive receives it in a usable form; the mechanisms differ in which institution that is. Figure 1. A five-stage conceptual framework connecting satellite observation, AI inference, and spatial-computing dissemination to regulatory engagement mechanisms and accountability conditions. Dashed arrows indicate feedback from accountability requirements to system design. Source: author’s elaboration based on the reviewed literature. Conditionality and Accountability of Algorithmic Monitoring RQ3 asks when monitoring changes outcomes and which accountability problems constrain it. The clearest finding in the verified literature is conditionality on institutional capacity. Moffette et al. (2021) found that subscriptions to near-real-time alerts across 22 tropical countries reduced the probability of deforestation by 18% in Africa relative to 2011 to 2016 averages, with stronger effects in protected areas and concessions, but detected no comparable effect on other continents and concluded that availability of alerts alone was insufficient. The experimental evidence points the same way: community monitoring in Peru shifted institutions more clearly than it shifted aggregate loss (Slough et al., 2021). Detection, in short, is a complement to enforcement capacity, not a substitute for it, which is consistent with the Brazilian evidence that the alert-enforcement coupling, not the alert, produces deterrence (Assunção et al., 2023). A second constraint is definitional mismatch. Satellites measure canopy disturbance; regulations govern legal categories such as deforestation for specific commodities after specific cutoff dates. Global maps of forest loss include harvest, fire, and natural disturbance alongside conversion, and driver attribution remains probabilistic (Curtis et al., 2018). Due-diligence regimes therefore require an inferential chain, from disturbance to conversion to commodity linkage, whose later links depend on land-use classification that is measurably harder than clearing detection (Masolele et al., 2021; Pendrill et al., 2022). Degradation compounds the mismatch, since processes responsible for carbon losses comparable to deforestation largely escape the alert systems on which regulatory attention concentrates (Bullock et al., 2020; Lapola et al., 2023). A third constraint is the distribution of error and burden. Zhunusova et al. (2022) argue that the compliance architecture of the EU regulation risks disadvantaging smallholders, indigenous peoples, and local communities in producer countries, who are least equipped to contest an adverse geospatial determination. The technical literature gives this concern empirical shape: radar alerting now resolves events smaller than half a hectare (Reiche et al., 2021), which brings smallholder clearing into regulatory view at exactly the moment when due-diligence obligations attach commercial consequences to detection. Accuracy statistics reported at system level do not describe the error experienced by any particular smallholder plot, and none of the verified platform studies reports a contestation or appeal channel for affected land users. Taken together, the answer to RQ3 is that algorithmic monitoring changes outcomes where an institution can act on it, mismeasures precisely the categories regulation cares about most, and currently externalizes its residual error onto the least resourced parties in the chain. 5. Discussion Theoretical Implications The review's conceptual contribution is to reposition deforestation mapping systems as regulatory intermediaries rather than as measurement instruments that happen to be used by regulators. The five-stage framework (Figure 1) makes this explicit: observation infrastructure, AI inference, platform dissemination, engagement mechanisms, and accountability conditions are distinct stages, each with stage-specific failure modes, and system performance is the product of the weakest stage rather than the sum of the strongest. This is an interpretive claim derived from the synthesis, not an empirical result, but it organizes otherwise disparate findings: the high classification accuracies of the detection literature (de Bem et al., 2020) coexist with null regulatory effects outside Africa (Moffette et al., 2021) because accuracy is a stage-two property while outcomes are decided at stages four and five. For governance theory, the framework specifies where the hardening of accountability described by Berning and Sotirov (2023) acquires its evidentiary base, and it suggests that transparency scholarship (Gardner et al., 2019) should treat platform design choices, such as those documented by Gomes et al. (2020), as governance variables. For remote-sensing science, it implies that reporting conventions built around reference-data accuracy systematically understate the properties regulators need, notably per-parcel error characterization and attributional confidence. The gap identified in the Introduction is closed to a definable extent. The review assembles the previously scattered evidence on the detection-to-engagement pathway into an explicit mechanism typology and framework, which is the integrative step the literature lacked. What the review cannot close, because the underlying studies do not yet exist, is the causal estimation of how specific technical properties, such as latency, minimum mapping unit, or attribution confidence, change regulatory outcomes across institutional settings; the framework specifies these as testable relationships rather than resolving them. Practical and Policy Implications For regulators implementing due-diligence legislation, the findings caution against treating any single map as adjudicative. Because satellites measure disturbance while the law regulates conversion for commodities, competent authorities need documented inferential procedures that combine alerts, post-clearing land-use classification, and plot-level geolocation, and they need explicit error tolerances before commercial sanctions attach. The concentration of illegality among identifiable properties (Rajão et al., 2020) supports risk-based verification rather than uniform screening. For national monitoring programmes, the Brazilian evidence indicates that the alert-to-enforcement coupling is the active ingredient (Assunção et al., 2023); programmes that replicate DETER's sensors without replicating its institutional receptor should not expect its results, a caution reinforced by the continent-scale heterogeneity in alert effectiveness (Moffette et al., 2021). For platform operators, the smallholder-visibility problem implies design duties: publishing per-alert confidence, supporting ground-truth feedback, and providing contestation channels would convert accountability from an aspiration into an interface property, in line with the stakeholder-engagement lessons of the early warning literature (Tabor & Holland, 2021) and the equity concerns raised for the EU regime (Zhunusova et al., 2022). For community-level programmes, the Peruvian experiment suggests that training and incentives, not data access alone, determine whether alerts become monitoring practice (Slough et al., 2021). Limitations The limitations follow from the method. First, an integrative review synthesizes published evaluations and cannot correct their imbalances: the evidence base is concentrated on Brazil, the Congo Basin, and pan-tropical platforms, so the framework's fit to South and Southeast Asian or dryland-forest contexts is less tested. Second, the review privileged English-language, DOI-registered journal literature and mandated institutional assessments; operational knowledge held in agency technical documents and non-English scholarship is under-represented, which may understate the sophistication of national programmes outside the documented cases. Third, no meta-analytic aggregation was attempted because outcome metrics are incommensurable across the constituent literatures, so comparative statements about method performance rest on individual studies rather than pooled estimates. Fourth, the framework derivation is interpretive; although each stage is anchored in at least two verified sources, other analysts could partition the pipeline differently. Finally, the regulatory environment is moving quickly, and analyses of the EUDR published before its full application necessarily assess design rather than realized enforcement practice. Future Research Three lines of work follow directly from what remains open. First, causal studies should treat technical properties as treatments: varying alert latency, spatial detail, or confidence presentation across comparable jurisdictions would identify which system properties actually move enforcement and compliance, extending the identification strategies of Assunção et al. (2023) and Moffette et al. (2021) from system presence to system design. Second, the measurement frontier should shift toward the categories regulation requires, especially degradation detection and post-clearing commodity attribution, where current capability is weakest relative to regulatory demand (Bullock et al., 2020; Masolele et al., 2021; Pendrill et al., 2022). Third, accountability research should examine algorithmic forest monitoring with the tools of regulatory scholarship, including empirical study of contestation, error redress, and the distributional incidence of false positives on smallholders, for which Zhunusova et al. (2022) provide the agenda-setting analysis. Progress on these three fronts would convert the framework proposed here from an organizing device into a tested theory of algorithmic environmental regulation. 6. Conclusion This article reviewed the evidence on AI and spatial computing for deforestation mapping and on the pathways through which mapped change informs regulation. The synthesis yields three conclusions. Detection capability is real and operational: time-series algorithms, deep learning, and cloud geospatial platforms now support disturbance alerting within days, at resolutions fine enough to see smallholder-scale clearing, across the humid tropics. Regulatory effect is conditional: alerts curb deforestation where coupled to enforcement capacity, as in Brazil and in African protected areas, and fail to do so where no institution can receive and act on them. Accountability is the unfinished stage: the categories satellites measure diverge from the categories law regulates, degradation largely escapes the systems that watch for clearing, and the burden of residual error falls on the least resourced actors. The five-stage framework developed here connects these conclusions and identifies the design and research choices, on latency, attribution, contestability, and institutional coupling, that will determine whether the next decade of algorithmic forest monitoring functions as regulatory infrastructure or merely as documentation of loss. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics. This study is a review of published literature and did not involve human participants, animal subjects, or personal data; ethical approval was therefore not required. Data Availability. No new data were created or analysed in this study. All sources synthesized are publicly available through the digital object identifiers listed in the references. References Assunção, J., Gandour, C., & Rocha, R. (2015). Deforestation slowdown in the Brazilian Amazon: Prices or policies? Environment and Development Economics, 20(6), 697–722. https://doi.org/10.1017/S1355770X15000078 Assunção, J., Gandour, C., & Rocha, R. (2023). DETERring deforestation in the Amazon: Environmental monitoring and law enforcement. American Economic Journal: Applied Economics, 15(2), 125–156. https://doi.org/10.1257/app.20200196 Bager, S. L., Persson, U. M., & dos Reis, T. N. P. (2021). Eighty-six EU policy options for reducing imported deforestation. One Earth, 4(2), 289–306. https://doi.org/10.1016/j.oneear.2021.01.011 Ball, J. G. C., Petrova, K., Coomes, D. A., & Flaxman, S. (2022). Using deep convolutional neural networks to forecast spatial patterns of Amazonian deforestation. Methods in Ecology and Evolution, 13(11), 2622–2634. https://doi.org/10.1111/2041-210X.13953 Berning, L., & Sotirov, M. (2023). Hardening corporate accountability in commodity supply chains under the European Union Deforestation Regulation. Regulation & Governance, 17(4), 870–890. https://doi.org/10.1111/rego.12540 Bullock, E. L., Woodcock, C. E., Souza, C., Jr., & Olofsson, P. (2020). Satellite-based estimates reveal widespread forest degradation in the Amazon. Global Change Biology, 26(5), 2956–2969. https://doi.org/10.1111/gcb.15029 Curtis, P. G., Slay, C. M., Harris, N. L., Tyukavina, A., & Hansen, M. C. (2018). Classifying drivers of global forest loss. Science, 361(6407), 1108–1111. https://doi.org/10.1126/science.aau3445 de Bem, P., de Carvalho Junior, O., Guimarães, R. F., & Gomes, R. T. (2020). Change detection of deforestation in the Brazilian Amazon using Landsat data and convolutional neural networks. Remote Sensing, 12(6), Article 901. https://doi.org/10.3390/rs12060901 Diniz, C. G., Souza, A. A. A., Santos, D. C., Dias, M. C., Luz, N. C., Moraes, D. R. V., Maia, J. S., Gomes, A. R., Narvaes, I. S., Valeriano, D. M., Maurano, L. E. P., & Adami, M. (2015). DETER-B: The new Amazon near real-time deforestation detection system. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(7), 3619–3628. https://doi.org/10.1109/JSTARS.2015.2437075 FAO. (2020). Global Forest Resources Assessment 2020. FAO. https://doi.org/10.4060/ca9825en Finer, M., Novoa, S., Weisse, M. J., Petersen, R., Mascaro, J., Souto, T., Stearns, F., & Martinez, R. G. (2018). Combating deforestation: From satellite to intervention. Science, 360(6395), 1303–1305. https://doi.org/10.1126/science.aat1203 Gardner, T. A., Benzie, M., Börner, J., Dawkins, E., Fick, S., Garrett, R., Godar, J., Grimard, A., Lake, S., Larsen, R. K., Mardas, N., McDermott, C. L., Meyfroidt, P., Osbeck, M., Persson, M., Sembres, T., Suavet, C., Strassburg, B., Trevisan, A., West, C., & Wolvekamp, P. (2019). Transparency and sustainability in global commodity supply chains. World Development, 121, 163–177. https://doi.org/10.1016/j.worlddev.2018.05.025 Gomes, V., Queiroz, G., & Ferreira, K. (2020). An overview of platforms for big Earth observation data management and analysis. Remote Sensing, 12(8), Article 1253. https://doi.org/10.3390/rs12081253 Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., & Moore, R. (2017). Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031 Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., Thau, D., Stehman, S. V., Goetz, S. J., Loveland, T. R., Kommareddy, A., Egorov, A., Chini, L., Justice, C. O., & Townshend, J. R. G. (2013). High-resolution global maps of 21st-century forest cover change. Science, 342(6160), 850–853. https://doi.org/10.1126/science.1244693 Hansen, M. C., Krylov, A., Tyukavina, A., Potapov, P. V., Turubanova, S., Zutta, B., Ifo, S., Margono, B., Stolle, F., & Moore, R. (2016). Humid tropical forest disturbance alerts using Landsat data. Environmental Research Letters, 11(3), Article 034008. https://doi.org/10.1088/1748-9326/11/3/034008 Hethcoat, M. G., Edwards, D. P., Carreiras, J. M. B., Bryant, R. G., França, F. M., & Quegan, S. (2019). A machine learning approach to map tropical selective logging. Remote Sensing of Environment, 221, 569–582. https://doi.org/10.1016/j.rse.2018.11.044 Lapola, D. M., Pinho, P., Barlow, J., Aragão, L. E. O. C., Berenguer, E., Carmenta, R., Liddy, H. M., Seixas, H., Silva, C. V. J., Silva-Junior, C. H. L., Alencar, A. A. C., Anderson, L. O., Armenteras, D., Brovkin, V., Calders, K., Chambers, J., Chini, L., Costa, M. H., Faria, B. L., ... Walker, W. S. (2023). The drivers and impacts of Amazon forest degradation. Science, 379(6630), Article eabp8622. https://doi.org/10.1126/science.abp8622 Ma, L., Liu, Y., Zhang, X., Ye, Y., Yin, G., & Johnson, B. A. (2019). Deep learning in remote sensing applications: A meta-analysis and review. ISPRS Journal of Photogrammetry and Remote Sensing, 152, 166–177. https://doi.org/10.1016/j.isprsjprs.2019.04.015 Masolele, R. N., De Sy, V., Herold, M., Marcos, D., Verbesselt, J., Gieseke, F., Mullissa, A. G., & Martius, C. (2021). Spatial and temporal deep learning methods for deriving land-use following deforestation: A pan-tropical case study using Landsat time series. Remote Sensing of Environment, 264, Article 112600. https://doi.org/10.1016/j.rse.2021.112600 Moffette, F., Alix-Garcia, J., Shea, K., & Pickens, A. H. (2021). The impact of near-real-time deforestation alerts across the tropics. Nature Climate Change, 11(2), 172–178. https://doi.org/10.1038/s41558-020-00956-w Nepstad, D., McGrath, D., Stickler, C., Alencar, A., Azevedo, A., Swette, B., Bezerra, T., DiGiano, M., Shimada, J., Seroa da Motta, R., Armijo, E., Castello, L., Brando, P., Hansen, M. C., McGrath-Horn, M., Carvalho, O., & Hess, L. (2014). Slowing Amazon deforestation through public policy and interventions in beef and soy supply chains. Science, 344(6188), 1118–1123. https://doi.org/10.1126/science.1248525 Pendrill, F., Gardner, T. A., Meyfroidt, P., Persson, U. M., Adams, J., Azevedo, T., Bastos Lima, M. G., Baumann, M., Curtis, P. G., De Sy, V., Garrett, R., Godar, J., Goldman, E. D., Hansen, M. C., Heilmayr, R., Herold, M., Kuemmerle, T., Lathuilière, M. J., Ribeiro, V., ... West, C. (2022). Disentangling the numbers behind agriculture-driven tropical deforestation. Science, 377(6611), Article eabm9267. https://doi.org/10.1126/science.abm9267 Rajão, R., Soares-Filho, B., Nunes, F., Börner, J., Machado, L., Assis, D., Oliveira, A., Pinto, L., Ribeiro, V., Rausch, L., Gibbs, H., & Figueira, D. (2020). The rotten apples of Brazil's agribusiness. Science, 369(6501), 246–248. https://doi.org/10.1126/science.aba6646 Reiche, J., Mullissa, A., Slagter, B., Gou, Y., Tsendbazar, N., Odongo-Braun, C., Vollrath, A., Weisse, M. J., Stolle, F., Pickens, A., Donchyts, G., Clinton, N., Gorelick, N., & Herold, M. (2021). Forest disturbance alerts for the Congo Basin using Sentinel-1. Environmental Research Letters, 16(2), Article 024005. https://doi.org/10.1088/1748-9326/abd0a8 Shekhar, S., Feiner, S. K., & Aref, W. G. (2015). Spatial computing. Communications of the ACM, 59(1), 72–81. https://doi.org/10.1145/2756547 Slough, T., Kopas, J., & Urpelainen, J. (2021). Satellite-based deforestation alerts with training and incentives for patrolling facilitate community monitoring in the Peruvian Amazon. Proceedings of the National Academy of Sciences, 118(29), Article e2015171118. https://doi.org/10.1073/pnas.2015171118 Tabor, K. M., & Holland, M. B. (2021). Opportunities for improving conservation early warning and alert systems. Remote Sensing in Ecology and Conservation, 7(1), 7–17. https://doi.org/10.1002/rse2.163 Tamiminia, H., Salehi, B., Mahdianpari, M., Quackenbush, L., Adeli, S., & Brisco, B. (2020). Google Earth Engine for geo-big data applications: A meta-analysis and systematic review. ISPRS Journal of Photogrammetry and Remote Sensing, 164, 152–170. https://doi.org/10.1016/j.isprsjprs.2020.04.001 Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367. https://doi.org/10.1177/1534484305278283 Turubanova, S., Potapov, P. V., Tyukavina, A., & Hansen, M. C. (2018). Ongoing primary forest loss in Brazil, Democratic Republic of the Congo, and Indonesia. Environmental Research Letters, 13(7), Article 074028. https://doi.org/10.1088/1748-9326/aacd1c Verbesselt, J., Zeileis, A., & Herold, M. (2012). Near real-time disturbance detection using satellite image time series. Remote Sensing of Environment, 123, 98–108. https://doi.org/10.1016/j.rse.2012.02.022 Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Wulder, M. A., Loveland, T. R., Roy, D. P., Crawford, C. J., Masek, J. G., Woodcock, C. E., Allen, R. G., Anderson, M. C., Belward, A. S., Cohen, W. B., Dwyer, J., Erb, A., Gao, F., Griffiths, P., Helder, D., Hermosilla, T., Hipple, J. D., Hostert, P., Hughes, M. J., ... Zhu, Z. (2019). Current status of Landsat program, science, and applications. Remote Sensing of Environment, 225, 127–147. https://doi.org/10.1016/j.rse.2019.02.015 Zhu, X. X., Tuia, D., Mou, L., Xia, G.-S., Zhang, L., Xu, F., & Fraundorfer, F. (2017). Deep learning in remote sensing: A comprehensive review and list of resources. IEEE Geoscience and Remote Sensing Magazine, 5(4), 8–36. https://doi.org/10.1109/MGRS.2017.2762307 Zhu, Z., & Woodcock, C. E. (2014). Continuous change detection and classification of land cover using all available Landsat data. Remote Sensing of Environment, 144, 152–171. https://doi.org/10.1016/j.rse.2014.01.011 Zhunusova, E., Ahimbisibwe, V., Sen, L. T. H., Sadeghi, A., Toledo-Aceves, T., Kabwe, G., & Günter, S. (2022). Potential impacts of the proposed EU regulation on deforestation-free supply chains on smallholders, indigenous peoples, and local communities in producer countries outside the EU. Forest Policy and Economics, 143, Article 102817. https://doi.org/10.1016/j.forpol.2022.102817 Hashtags: #SDG15 #LifeOnLand #SDG13 #ClimateAction #SDG12 #ResponsibleConsumption #DeforestationMonitoring #ArtificialIntelligence #SpatialComputing #RemoteSensing #EnvironmentalGovernance #DeepLearning #SupplyChainRegulation #U7YJournal #U7Y #AcademicResearch

Author: Emma Dubois Affiliation: Swiss International University (SIU) ORCID ID: 0009-0001-8464-158X Submitted 05 April 2026; Revised 02 June 2026; Accepted 17 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10001 Volume 3, December 2026, (SpecSDG10001) Abstract Cross-border academic partnerships now carry a substantial share of the world's internationally delivered higher education, yet the rules that assure their quality have been written in several places at once: intergovernmental guidelines, regional standards, national accreditation regimes, and the informal discipline of rankings and benchmarking. This article presents an integrative review of peer-reviewed scholarship and documented policy instruments published between 2005 and 2026, combined into a conceptual account of how formal quality assurance standards and comparative benchmarking jointly steer partnership strategy. The review traces the evolution of the principal instruments, from the OECD and UNESCO guidelines of 2005 through the European Standards and Guidelines, the European Approach for Quality Assurance of Joint Programmes, the UNESCO Global Convention on recognition, and the INQAAHE international standards, and reads them against the management literature on branch campuses, franchised provision, and joint programmes. Three findings organise the argument. First, formal standards have moved from consumer protection toward trust-based, mode-neutral frameworks that allocate responsibility between sending and receiving systems. Second, benchmarking and rankings operate as a parallel, non-mandated governance channel that shapes partner selection and institutional behaviour while measuring partnership quality only weakly. Third, national regimes translate shared standards selectively, so partnerships face layered and sometimes duplicated requirements. The article contributes a multi-level framework linking partnership strategy, standards, and benchmarking, and identifies implications for institutions, agencies, and policy makers. Keywords: transnational higher education, quality assurance, cross-border partnerships, benchmarking, accreditation, international branch campuses, joint degrees 1. Introduction Universities increasingly pursue their international ambitions through partnerships that cross jurisdictions: franchised programmes, jointly awarded degrees, co-founded institutions, and branch campuses established on foreign soil. These arrangements have grown from a peripheral activity into a recognised mode of provision with its own vocabulary, data problems, and regulatory politics (Knight, 2016). Their scale is no longer trivial. Reviews of the branch campus sector counted more than two hundred such campuses worldwide by the mid-2010s, concentrated in the Middle East and South-east Asia (Healey, 2015), and one systematic mapping recorded a peak of 249 campuses across thirty-three countries in January 2017 (Escrivá-Beltran et al., 2019). In the United Kingdom, the largest sending system, roughly 700,000 students were studying for UK qualifications in their own countries in 2017/18, more than twice the number of non-European Union students who travelled to the UK to study (Healey, 2020). Growth of this kind places unusual demands on quality assurance. A degree taught in one country under the academic authority of an institution in another sits at the intersection of two regulatory systems, two labour markets, and often two sets of accreditation requirements. The instruments designed to govern this intersection have multiplied since the mid-2000s: the OECD and UNESCO issued joint guidelines on cross-border provision in 2005 (OECD & UNESCO, 2005); the European Higher Education Area revised its Standards and Guidelines in 2015 and adopted a dedicated European Approach for the external evaluation of joint programmes in the same year (EHEA, 2015; ENQA et al., 2015); UNESCO's Global Convention on the recognition of higher education qualifications was adopted in 2019 and entered into force in 2023 (UNESCO, 2019); and the international network of quality assurance agencies issued global standards for external quality assurance providers, including a module on cross-border activity, in 2022 (INQAAHE, 2022). Alongside this formal architecture, a second and less orderly set of comparative instruments has come to influence the same decisions. Global rankings and institutional benchmarking shape which partners institutions consider credible, how governments frame excellence, and how quality itself is argued about in public (Hazelkorn, 2015; Marginson, 2014; Musselin, 2018). The scholarly literatures that examine these developments have grown up largely in separate rooms. Research on transnational education has been reviewed as a field in its own right, with well-mapped clusters on management, pedagogy, and policy (Kosmützky & Putty, 2016), and the branch campus literature has been synthesised repeatedly from an institutional and stakeholder perspective (Escrivá-Beltran et al., 2019; Wilkins, 2021). Quality assurance scholarship, for its part, has concentrated on domestic systems: how agencies translate shared European standards into national frameworks (Manatos & Huisman, 2020), what makes internal quality management effective (Seyfried & Pohlenz, 2018), and how far evidence can support quality assurance policy at all (Beerkens, 2018). The rankings literature has analysed comparative instruments as policy discourse and as competition infrastructure (Erkkilä, 2014; Musselin, 2018), but rarely in connection with the quality assurance of partnerships. What the literature has not resolved is how these two governance channels, formal standards and comparative benchmarking, interact when they bear on the same object: the strategic cross-border partnership. This gap matters for institutional leaders who must satisfy layered accreditation requirements while competing for reputationally attractive partners, for quality assurance agencies asked to review provision delivered far from their jurisdiction, and for policy makers deciding whether recognition and quality instruments should converge globally or remain regional. This article addresses that gap through an integrative review with a conceptual contribution. It asks three questions. First, how have formal quality assurance standards and instruments for cross-border provision evolved since 2005, and through what mechanisms do they govern partnerships (RQ1)? Second, how does benchmarking, including global rankings, operate as a parallel governance instrument shaping partnership strategy and quality practice (RQ2)? Third, what quality assurance demands do different partnership models generate, and where do the resulting arrangements converge or diverge across institutional, national, and supranational levels (RQ3)? The contribution is twofold: a synthesis that reads the policy instruments and the management scholarship against each other, and a conceptual framework that locates partnership strategy within the joint operation of standards-based and benchmarking-based steering. The article proceeds as follows. The next section situates the review in the literatures on internationalisation, quality assurance, and comparative instruments, and sharpens the gap. The Method section sets out the review protocol. The Findings section develops four themes: the evolution of formal standards, benchmarking as governance, the quality assurance demands of partnership models, and convergence and divergence across levels. The Discussion draws theoretical and practical implications, states the limitations of the approach, and identifies the questions that remain open. 2. Background and Conceptual Orientation 2.1. Internationalisation and the Rise of Partnership-Based Provision Internationalisation is best understood as the set of policies and practices through which academic systems and institutions respond to a global academic environment, driven by motives that range from commercial advantage to knowledge acquisition and curriculum enrichment (Altbach & Knight, 2007). Over five decades it has moved from the margins of institutional life to the centre of reform agendas, and its instruments have diversified accordingly (de Wit & Altbach, 2021). Within this broader movement, the mobility of programmes and providers, rather than students, constitutes a distinct strand. Knight (2016) argues that this strand changed dramatically in scope and scale during the 2010s, generating new actors, new partnership forms, and a proliferation of overlapping terms whose confusion carries real costs for quality assurance, enrolment planning, and regulation. The classification developed for the British Council and the German Academic Exchange Service brings order to this variety by distinguishing independent provision, in which the sending institution retains responsibility for design, delivery, and external quality assurance, from collaborative provision, in which sending and host institutions share those responsibilities, and by sorting programme and provider mobility into six categories on that basis (Knight & McNamara, 2017). Bibliometric work confirms that scholarship followed practice. Kosmützky and Putty (2016) show that research on transnational, offshore, cross-border, and borderless higher education consolidated into a recognisable thematic field from the early 2000s, with identifiable topic clusters and citation structures. The branch campus, as the most capital-intensive partnership-adjacent form, has attracted particular attention: institutional-theory accounts explain campus establishment through considerations of legitimacy, status, institutional distance, risk, and revenue (Wilkins & Huisman, 2012), while management-oriented reviews emphasise the difficulty of balancing stakeholders that include home-country quality regulators, host governments, and local joint-venture partners (Healey, 2015). Two decades of development have produced both expansion and failure, and continued viability appears to depend on sustained benefit to students, institutions, and host governments alike (Wilkins, 2021). 2.2. Quality Assurance as a Governance Field Quality assurance occupies an uneasy position between improvement and control. Empirical work on its effectiveness suggests that outcomes depend heavily on organisational conditions: quality managers perceive their work as effective where institutional leadership supports it and where institutions cooperate with one another, and as ineffective where quality assurance is experienced as sanction or administrative burden (Seyfried & Pohlenz, 2018). At the policy level, Beerkens (2018) argues that rigorous ex-post impact evaluation of quality assurance is exceptionally difficult, and that policy should instead cultivate more realistic ways for evidence to inform quality work. These findings caution against treating any standards regime, national or supranational, as self-executing. The supranational layer has nonetheless thickened. Zapp and Ramirez (2019) document the construction of a global higher education regime along discursive, normative, and regulatory dimensions: a growing network of international organisations, a striking increase in the number of national accreditation agencies and in their mutual cross-national recognition, and parallel growth in qualifications frameworks and regional recognition conventions. Within Europe, the Standards and Guidelines for Quality Assurance in the European Higher Education Area function as the shared reference point for internal quality assurance, external review, and the registration of agencies (ENQA et al., 2015). Yet shared standards are not uniformly applied. Manatos and Huisman (2020) find, across review documentation covering seventeen institutions in four countries, that national quality assurance frameworks often deviate from the European standards through processes of copying and modification, even while review panels stay close to their national frameworks. Translation, not transposition, is the operative mechanism. 2.3. Rankings and Benchmarking as Comparative Instruments A separate literature treats comparison itself as a governance technology. Hazelkorn (2015) shows that global rankings influence institutional strategy, student choice, and national policy well beyond what their methods warrant. Marginson (2014) evaluates the main ranking systems against social science criteria and concludes that several rest on reputational recycling rather than observable performance, which limits their value as public information. Erkkilä (2014) traces how rankings constructed a policy problem of European higher education, fed convergent reform agendas across countries, and produced stratification and homogenisation as side effects. Musselin (2018) generalises the point: competition in higher education has become institutional and multi-level, organised through impersonal judgment devices, and it generates new classifications and alliances among institutions that compete and cooperate at once. Benchmarking in the narrower managerial sense, structured comparison against peers for improvement, has meanwhile been studied as a quality tool in its own right, with case-based evidence that comparison and assessment can support institutional improvement agendas (Tasopoulou & Tsiotras, 2017). Evidence on the internationalisation indicators inside rankings, however, indicates that their quantitative orientation makes them a poor measure of the quality of international activity (Hauptman Komotar, 2019). Read together, these literatures leave a specific question unanswered. The transnational education scholarship analyses partnership forms and their management; the quality assurance scholarship analyses standards and their translation; the rankings scholarship analyses comparison and its effects. No study known to this review has integrated the three to explain how formal standards and comparative benchmarking jointly govern strategic cross-border partnerships across institutional, national, and supranational levels. That integration is the task of this article. 3. Method 3.1. Design and Rationale The study is an integrative review with a conceptual contribution. The integrative review is the appropriate design where the relevant evidence is heterogeneous in method and genre, because it is the review form that permits the combination of empirical and theoretical sources within a single analysis (Whittemore & Knafl, 2005). It is also the form best suited to producing new conceptual structure rather than a pooled effect estimate (Torraco, 2016). A systematic review in the strict sense was rejected because the question spans policy instruments that are not indexed in scholarly databases, and a purely conceptual paper was rejected because the argument depends on the accumulated empirical record. The review followed the staged logic recommended for integrative work: problem identification, literature search, evaluation, analysis, and presentation (Whittemore & Knafl, 2005), with reporting conventions informed by general guidance on review methodology (Snyder, 2019). 3.2. Search Strategy and Selection of Scholarship Scholarly sources were identified through structured searches of Scopus, Web of Science, and Google Scholar conducted in 2026, covering publications from January 2005 to August 2026, with emphasis on work published from 2016 onward. Example search strings, adapted to each database's syntax, included: ("transnational education" OR "cross-border higher education" OR "international branch campus" OR "joint degree" OR "franchised programme") AND ("quality assurance" OR accreditation OR standards); and (benchmarking OR rankings) AND ("higher education" AND (governance OR internationali*)). Reference lists of retrieved reviews were hand-searched for additional sources. Inclusion criteria were: (a) peer-reviewed journal articles, research monographs, or edited-volume chapters; (b) direct relevance to at least one of the three research questions, that is, to the quality assurance of cross-border provision, to partnership forms and their management, or to benchmarking and rankings as governance instruments; (c) English language; and (d) conceptual, empirical, or review contributions with an identifiable method or explicit theoretical argument. Exclusion criteria were: (a) purely descriptive practice reports without analytical content; (b) studies of student mobility without a provision or partnership dimension; (c) national case studies with no implication beyond their setting; and (d) sources whose bibliographic details could not be verified against the publisher's or an indexing service's record. Screening proceeded in two stages, title and abstract followed by full text, applying the criteria in that order and retaining methodological anchor works on review method regardless of topic. The final scholarly corpus comprises twenty-six peer-reviewed items spanning higher education studies, comparative education, management, and policy analysis. 3.3. Selection and Treatment of Policy Instruments Because formal standards are constitutive of the object under review, six documented policy instruments were analysed alongside the scholarship: the OECD and UNESCO guidelines on cross-border provision (OECD & UNESCO, 2005); the European Standards and Guidelines (ENQA et al., 2015); the European Approach for Quality Assurance of Joint Programmes (EHEA, 2015); the transnational education classification framework commissioned by the British Council and DAAD (Knight & McNamara, 2017); the UNESCO Global Convention on recognition (UNESCO, 2019); and the INQAAHE international standards and guidelines (INQAAHE, 2022). Instruments were selected on three criteria: issuance by an intergovernmental body, a ministerial conference, or a recognised international network; explicit applicability to provision or quality assurance that crosses borders; and documented adoption or entry into force. Each instrument was read in its official published version and coded for scope, addressees, mechanism (binding, voluntary, or standard-setting), and treatment of cross-border provision. 3.4. Analysis and Framework Derivation Analysis followed a thematic synthesis in three passes. The first pass coded each source against the research questions. The second pass grouped codes into candidate themes and tested them against disconfirming sources; for example, claims of global convergence drawn from regime-level scholarship were checked against comparative evidence of national divergence. The third pass arranged the surviving themes into the four findings reported below. The conceptual framework was derived abductively: the levels (supranational, national, institutional) were taken from the structure of the instruments themselves, the two steering channels (standards and benchmarking) from the juxtaposition of the quality assurance and rankings literatures, and the feedback relations from the empirical partnership studies. The framework is interpretive; it organises the reviewed evidence and is not itself a tested model. 3.5. Rigour and Trustworthiness Several measures limited bias. Every bibliographic record and every quantitative claim carried into the synthesis was verified against the publisher's record, an indexing service, or the issuing organisation's own publication before inclusion, and sources that could not be verified were discarded. Claims are attributed at the strength the underlying study supports, and single-study findings are not presented as consensus. The scope of the review is bounded: it privileges English-language scholarship, which over-represents the perspectives of major sending systems, and it treats policy instruments as texts, not as implemented practice. These boundaries are revisited in the Limitations subsection. 4. Findings 4.1. The Evolution of Formal QA Standards for Cross-Border Provision The formal architecture governing cross-border provision has passed through three recognisable phases, summarised in Table 1. The first phase was protective. The 2005 guidelines issued jointly by the OECD and UNESCO responded to the rapid growth of provision that escaped national oversight, offering an international framework intended to protect students and other stakeholders from low-quality provision and disreputable providers, and addressing themselves to governments, institutions, quality assurance agencies, and recognition bodies in both sending and receiving countries (OECD & UNESCO, 2005). Their mechanism was voluntary guidance organised around the principle of shared responsibility; they created expectations rather than obligations. The second phase was regional and trust-based. The 2015 revision of the European Standards and Guidelines, adopted by ministers of the European Higher Education Area at Yerevan, established a mode-neutral principle with direct consequences for partnerships: the standards are designed to apply to all higher education, regardless of place or mode of delivery (ENQA et al., 2015). A degree franchised abroad or taught at a branch campus therefore falls within the same internal and external quality assurance expectations as home provision. In the same year, ministers adopted the European Approach for Quality Assurance of Joint Programmes, which defines standards drawn from agreed European tools and is designed to be used without additional national criteria, so that a joint programme can undergo a single external evaluation rather than parallel accreditations in every partner country (EHEA, 2015). The European Approach is the clearest institutional acknowledgement that partnership-based provision breaks the assumptions of nationally contained quality assurance, and its design principle, one evaluation accepted by many systems, is a template other regions have yet to replicate. The third phase is global and infrastructural. The classification framework commissioned by the British Council and DAAD supplies the definitional layer that data collection and regulation require, distinguishing collaborative from independent provision and sorting programme and provider mobility into six categories according to who awards the qualification, who designs the curriculum, and who conducts external quality assurance (Knight & McNamara, 2017). The UNESCO Global Convention, adopted in 2019 and in force since March 2023, extends recognition obligations to the global level as the first worldwide legal framework for fair, transparent, and non-discriminatory recognition of qualifications, complementing the existing regional conventions; by 2026 it had been ratified by forty-one states (UNESCO, 2019). The INQAAHE international standards of 2022 complete the picture from the agency side: they set baseline standards for external quality assurance providers across tertiary education and include a dedicated module for cross-border quality assurance and for the quality assurance of provision delivered outside the jurisdiction in which a provider is legally incorporated (INQAAHE, 2022). Answering RQ1, the trajectory runs from consumer protection through regional mutual trust toward a global infrastructure of definitions, recognition law, and agency standards, with the mechanism shifting from exhortation toward standard-setting embedded in registration, recognition, and treaty instruments. Table 1 International Instruments Governing the Quality Assurance of Cross-Border Higher Education Provision Instrument (year) Issuing body Scope and addressees Mechanism Relevance to cross-border partnerships Guidelines for Quality Provision in Cross-border Higher Education (2005) OECD and UNESCO Global; governments, institutions, QA agencies, recognition bodies in sending and receiving countries Voluntary guidelines; shared responsibility between systems First international framework aimed at protecting students from low-quality cross-border provision (OECD & UNESCO, 2005) Standards and Guidelines for QA in the EHEA, ESG (2015) E4 group; adopted by EHEA ministers, Yerevan EHEA; all higher education regardless of place or mode of delivery Standard-setting in three parts (internal QA, external QA, agencies); basis for agency registration Extends home quality assurance expectations to provision delivered abroad (ENQA et al., 2015) European Approach for QA of Joint Programmes (2015) EHEA ministers Joint programmes within the EHEA Standards based on agreed EHEA tools, applied without additional national criteria Enables a single external evaluation of a joint programme across partner systems (EHEA, 2015) TNE Classification Framework (2017) British Council and DAAD Global; programme and provider mobility in six categories Common definitions; responsibility sorted by award, curriculum design, and external QA Distinguishes independent from collaborative provision and clarifies which partner carries QA obligations (Knight & McNamara, 2017) Global Convention on the Recognition of Qualifications (2019) UNESCO Global; states parties (41 ratifications by 2026) Binding treaty; entered into force 2023; complements five regional conventions First worldwide legal framework for fair and transparent recognition of qualifications earned across borders (UNESCO, 2019) International Standards and Guidelines, ISG (2022) INQAAHE Global; external QA providers across tertiary education (ISCED 4-8) Baseline standards plus selective modules Dedicated module on cross-border QA and on QA of provision delivered outside the provider’s jurisdiction (INQAAHE, 2022) Note. Compiled from the adopted texts of the instruments as cited in each row; entries summarise scope and mechanism as stated in the official published versions. 4.2. Benchmarking as Governance: The Second Steering Channel Formal standards are not the only comparative pressure bearing on partnerships. Rankings and benchmarking constitute a second channel that operates without mandate but with considerable force. Hazelkorn (2015) documents how rankings shape institutional strategy, government policy, and student decisions across systems. Erkkilä (2014) shows that in Europe rankings did more than describe: they constructed a policy problem, positioned higher education as an element of economic competitiveness, and fed convergent reform agendas whose side effects include stratification and homogenisation. Musselin (2018) provides the structural reading: competition among universities has become institutional and multi-level, organised through impersonal judgment devices such as rankings and classifications, and it simultaneously produces new alliances among institutions that recognise one another as members of the same category. This second channel intersects with partnership strategy at several points. The institutional-theory literature on branch campuses identifies status and legitimacy among the principal motives for establishing a foreign presence (Wilkins & Huisman, 2012), and the alliance-formation dynamic that Musselin (2018) describes implies that ranked position conditions who is considered an eligible partner at all. Benchmarking in its managerial form offers a more constructive face: case-based evidence indicates that structured comparison against peers can support quality improvement in universities committed to it (Tasopoulou & Tsiotras, 2017). Yet the measurement base of the comparative channel is weak precisely where partnerships are concerned. Marginson (2014) evaluates the major ranking systems against criteria of materiality, objectivity, and externality and argues that rankings which recycle reputation without connection to observable outputs offer no common value, and Hauptman Komotar (2019) concludes that the internationalisation indicators inside global rankings are too crude to serve as measures of the quality of international activity. The answer this review gives to RQ2 is therefore double-edged: benchmarking steers partnership behaviour, through partner selection, government expectations, and the framing of excellence, while measuring the quality of partnership activity poorly. Institutions respond to a signal that is consequential but only loosely coupled to the educational quality that formal standards address. 4.3. Partnership Models and Their QA Demands The quality assurance burden a partnership carries depends on its form. The classification distinguishing independent from collaborative provision makes the point structurally: where a sending institution franchises a programme or operates a branch campus, responsibility for design, delivery, and external quality assurance remains primarily with the sender, whereas partnership programmes, joint universities, and jointly awarded degrees distribute those responsibilities across institutions and therefore across regulatory systems (Knight & McNamara, 2017). Each configuration generates a characteristic demand profile. Independent forms concentrate risk in the sending institution. Branch campuses expose their parents to financial and reputational hazard, and their managers must satisfy home-country quality regulators and host-country authorities simultaneously while answering to local joint-venture partners (Healey, 2015). The strategic literature describes this as dual embeddedness: campuses are pressed to conform to host-country institutions and, at the same time, to preserve the parent's identity and standards. Comparative evidence from Australian and British operations in South-east Asia indicates that campuses resolve this tension unevenly, maintaining close similarity to the parent in curriculum while diverging in staffing, which emerges as the most persistent strategic difficulty (Shams & Huisman, 2016). Because curriculum fidelity is what home quality assurance typically audits, the components most visible to external review may be the best controlled, while the staffing dimension on which teaching quality depends is the hardest to hold to parent standards. Collaborative forms distribute risk but multiply interfaces. Joint programmes answer to as many external quality assurance regimes as they have degree-awarding partners, which is the duplication the European Approach was designed to remove (EHEA, 2015). Below the level of formal compliance, partnership quality is produced operationally. Bordogna (2018) argues that the transnational education discourse has concentrated on strategic implementation, quality assurance, and pedagogy while neglecting the social relationships among the faculty members who actually deliver collaborative provision, and proposes that partnership development is strengthened where those operational relationships are understood and cultivated. Sustainability pressures cut across both forms: reviews of two decades of branch campus development record failures alongside expansions and conclude that provision persists only while it benefits students, institutions, and host countries at once (Wilkins, 2021), and analysis of the UK sector argues that the high-water mark of transnational enrolment growth may have passed, with growing scepticism about net economic benefit (Healey, 2020). Answering RQ3 in its first part: independent forms demand the extension of home quality systems across borders, collaborative forms demand the coordination of multiple external regimes, and both depend on operational and relational conditions that formal review reaches only indirectly. 4.4. Convergence, Divergence, and the Multi-Level Framework The final theme concerns how the levels fit together. At the global level, the evidence points to convergence in form. Zapp and Ramirez (2019) document a thickening regime: more international organisations active in higher education, a striking increase in national accreditation agencies and in their mutual recognition, and proliferating qualifications frameworks and recognition conventions. At the level of implementation, however, divergence persists. Manatos and Huisman (2020) show that even within the European Higher Education Area, national frameworks deviate from the shared standards through translation processes of copying and modification. Outside Europe the spread is wider still. Comparative analysis of branch campus regulation in Malaysia, Singapore, China, and South Korea finds that although all four systems expect internal quality mechanisms within campuses, their external approaches split: Singapore and South Korea regulate liberally and rely largely on home-country accreditation, while Malaysia and China subject foreign provision to comprehensive domestic quality assurance processes (Hou et al., 2018). Regional syntheses of Asian practice confirm the variety of arrangements under which transnational provision is assured (Hou et al., 2021). Convergence in form therefore coexists with divergence in practice, and partnerships inhabit the space between: a joint programme or branch campus may face duplicated review, or a regulatory gap, depending on how the systems it spans have translated the common instruments. Figure 1 integrates the four themes into a conceptual framework. Three levels are distinguished. At the supranational level sit the formal instruments (guidelines, standards, conventions, agency standards) and the comparative infrastructure (rankings, classifications, benchmarking exercises). At the national level, quality assurance regimes translate the formal instruments selectively, while governments import comparative vocabulary into policy targets. At the institutional level, partnership strategy responds to both channels: standards define what a partnership must demonstrate, benchmarking shapes which partnerships are sought and how they are justified. The two channels differ in mechanism, standards operating through review, registration, and recognition, benchmarking through reputation and market signal, but they meet in the partnership, which must satisfy the first while competing on the second. Feedback runs upward as well: partnership practice exposes gaps in standards, as joint programmes did in Europe before 2015 (EHEA, 2015), and institutional competition supplies rankings with their data and their audience (Musselin, 2018). The framework is interpretive rather than predictive; its claim is that neither channel can be understood in isolation when the object of governance is a cross-border partnership. Figure 1. Two-channel, multi-level framework of quality governance for cross-border academic partnerships. Solid arrows denote steering; dashed arrows denote feedback. Source: author’s elaboration of the reviewed literature. 5. Discussion 5.1. Theoretical Implications The review's central theoretical claim is that cross-border partnership governance is bi-channel. Scholarship that treats quality assurance as the whole story misses the extent to which comparative instruments condition partner choice and strategic justification (Hazelkorn, 2015; Musselin, 2018); scholarship that treats rankings as the whole story misses the dense standards infrastructure that determines whether a partnership can operate at all (ENQA et al., 2015; INQAAHE, 2022). Placing the two channels in one frame also refines the convergence debate. The global regime described by Zapp and Ramirez (2019) is real at the level of organisations and instruments, but the translation findings of Manatos and Huisman (2020) and the regulatory contrasts documented by Hou et al. (2018) show that convergence thins as it approaches practice. For partnerships this is not an abstract point: the layering of incompletely converged systems is precisely what generates duplicated accreditation, and instruments such as the European Approach are best read as targeted repairs to that layering (EHEA, 2015). The framework further suggests a re-reading of the branch campus literature: dual embeddedness (Shams & Huisman, 2016) is not only an organisational condition but a quality assurance condition, since the two systems in which a campus is embedded each carry their own review expectations. Against the gap identified in the Introduction, the review closes the integration deficit at the conceptual level: it specifies the levels, channels, and feedback relations through which standards and benchmarking jointly bear on partnerships, and it grounds each relation in verified literature. What it cannot close is the empirical deficit. The framework has not been tested against primary data, and the strength of the two channels relative to each other remains an open empirical question. 5.2. Implications for Institutions and QA Agencies For institutional leaders, the practical lesson is to treat the two channels explicitly and separately in partnership strategy. Reputational attractiveness, the benchmarking channel, is a legitimate selection criterion but a poor proxy for the quality of what a partner will deliver, given the weak measurement base of the comparative instruments (Hauptman Komotar, 2019; Marginson, 2014). Due diligence should therefore interrogate the standards channel directly: which external regimes will the partnership answer to, whether the systems involved accept single evaluations of joint provision, and how curriculum and staffing quality will be held to parent standards across borders, the dimension the empirical literature identifies as hardest to control (Shams & Huisman, 2016). Attention to the operational level is equally supported: partnerships develop through the working relationships of the faculty who deliver them, not only through the agreements that constitute them (Bordogna, 2018). For quality assurance agencies, two implications follow. First, the effectiveness literature indicates that quality assurance works where it is supported and cooperative rather than punitive (Seyfried & Pohlenz, 2018), which argues for review models of cross-border provision built on inter-agency cooperation rather than duplicated inspection; the INQAAHE module on cross-border activity and the European Approach both point in this direction (EHEA, 2015; INQAAHE, 2022). Second, the caution of Beerkens (2018) about impact evidence applies with extra force across borders, where outcome attribution is even harder; agencies should be modest about what review can demonstrate and invest in the information exchange that recognition instruments now require (UNESCO, 2019). For policy makers, the classification framework's allocation of responsibility by who awards, who designs, and who assures (Knight & McNamara, 2017) offers a workable basis for closing the regulatory gaps that liberal host regimes can leave open (Hou et al., 2018). 5.3. Limitations The limitations follow from the method. First, the corpus is English-language and skews toward major sending systems and toward Europe, where the instrument set is densest; quality assurance practice in receiving systems outside the cases reviewed is under-represented. Second, policy instruments were analysed as adopted texts, and the review can therefore say little about enforcement or lived compliance beyond what the cited implementation studies report. Third, the integrative design synthesises heterogeneous evidence and does not weight studies by design quality in the manner of a meta-analysis; the findings are a structured interpretation, not a pooled estimate. Fourth, the framework derived here is conceptual and untested, and its two-channel structure may understate other steering mechanisms, such as professional accreditation or funding conditionality, that were outside the review's scope. Finally, the field moves quickly; instruments adopted or revised after the search window closed in 2026 are not reflected. 5.4. Future Research Three lines of inquiry follow directly from what remains open. First, the relative force of the two channels is untested: comparative case studies of partnership formation could examine whether ranking position or accreditation status better predicts partner selection and survival. Second, implementation research on the newer global instruments is needed, particularly on whether the UNESCO convention's information-exchange obligations change agency behaviour (UNESCO, 2019) and whether the INQAAHE cross-border module shifts review practice outside Europe (INQAAHE, 2022). Third, the staffing dimension identified as the persistent weak point of cross-border quality (Shams & Huisman, 2016) deserves study connecting employment conditions at partner sites to the quality judgments of external review. Methodologically, the field would benefit from the common data infrastructure that the classification framework was designed to enable (Knight, 2016; Knight & McNamara, 2017), without which claims about growth and decline rest on national statistics of uneven coverage. 6. Conclusion This review set out to explain how the quality of strategic cross-border academic partnerships is governed. The evidence supports three conclusions. Formal quality assurance standards for cross-border provision have evolved from voluntary consumer protection into a layered infrastructure of mode-neutral standards, single-evaluation procedures, recognition law, and agency standards. Benchmarking and rankings form a second governance channel that shapes partnership strategy through reputation and competition while measuring partnership quality only weakly. And the demands a partnership faces depend jointly on its form, independent or collaborative, and on how the national systems it spans have translated the common instruments, so that formal convergence coexists with practical divergence. The conceptual framework developed here holds these elements in a single multi-level structure and offers institutions, agencies, and researchers a shared map of a governance space that has, until now, been studied one channel at a time. Declarations Funding: This research received no external funding. Conflicts of Interest: The author declares no conflict of interest. Ethics: This study did not involve human participants, animals, or personal data; ethical approval was therefore not required. Data Availability: No new data were created or analysed in this study. All sources reviewed are publicly available and cited in the reference list. References Altbach, P. G., & Knight, J. (2007). The internationalization of higher education: Motivations and realities. Journal of Studies in International Education, 11(3–4), 290–305. https://doi.org/10.1177/1028315307303542 Beerkens, M. (2018). Evidence-based policy and higher education quality assurance: Progress, pitfalls and promise. European Journal of Higher Education, 8(3), 272–287. https://doi.org/10.1080/21568235.2018.1475248 Bordogna, C. M. (2018). Transnational higher education partnerships and the role of operational faculty members: Developing an alternative theoretical approach for empirical research. Journal of Studies in International Education, 22(1), 3–19. https://doi.org/10.1177/1028315317724555 de Wit, H., & Altbach, P. G. (2021). Internationalization in higher education: Global trends and recommendations for its future. Policy Reviews in Higher Education, 5(1), 28–46. https://doi.org/10.1080/23322969.2020.1820898 Erkkilä, T. (2014). Global university rankings, transnational policy discourse and higher education in Europe. European Journal of Education, 49(1), 91–101. https://doi.org/10.1111/ejed.12063 Escrivá-Beltran, M., Muñoz-de-Prat, J., & Villó, C. (2019). Insights into international branch campuses: Mapping trends through a systematic review. Journal of Business Research, 101, 507–515. https://doi.org/10.1016/j.jbusres.2018.12.049 European Association for Quality Assurance in Higher Education, European Students' Union, European University Association, & European Association of Institutions in Higher Education. (2015). Standards and guidelines for quality assurance in the European Higher Education Area (ESG). https://www.enqa.eu/wp-content/uploads/2015/11/ESG_2015.pdf European Higher Education Area. (2015). European approach for quality assurance of joint programmes. https://www.eqar.eu/kb/joint-programmes/ Hauptman Komotar, M. (2019). Global university rankings and their impact on the internationalisation of higher education. European Journal of Education, 54(2), 299–310. https://doi.org/10.1111/ejed.12332 Hazelkorn, E. (2015). Rankings and the reshaping of higher education: The battle for world-class excellence (2nd ed.). Palgrave Macmillan. https://doi.org/10.1057/9781137446671 Healey, N. M. (2015). Managing international branch campuses: What do we know? Higher Education Quarterly, 69(4), 386–409. https://doi.org/10.1111/hequ.12082 Healey, N. M. (2020). The end of transnational education? The view from the UK. Perspectives: Policy and Practice in Higher Education, 24(3), 102–112. https://doi.org/10.1080/13603108.2019.1631227 Hou, A. Y. C., Hill, C., Chen, K. H. J., & Tsai, S. (2018). A comparative study of international branch campuses in Malaysia, Singapore, China, and South Korea: Regulation, governance, and quality assurance. Asia Pacific Education Review, 19(4), 543–555. https://doi.org/10.1007/s12564-018-9550-9 Hou, A. Y. C., Hsu, P. C., Chou, H. C., & Lin, A. F. Y. (2021). Quality assurance of transnational education in Asia. In C. Hill, J. Lamie, & T. Gore (Eds.), The evolution of transnational education: Pathways, globalisation and emerging trends (pp. 137–151). Routledge. https://doi.org/10.4324/9781003225386-14 International Network for Quality Assurance Agencies in Higher Education. (2022). International standards and guidelines for quality assurance in tertiary education (ISG). https://www.inqaahe.org/wp-content/uploads/2024/05/INQAAHE-International-Standards-and-Guidelines-ISG.pdf Knight, J. (2016). Transnational education remodeled: Toward a common TNE framework and definitions. Journal of Studies in International Education, 20(1), 34–47. https://doi.org/10.1177/1028315315602927 Knight, J., & McNamara, J. (2017). Transnational education: A classification framework and data collection guidelines for international programme and provider mobility (IPPM). British Council & DAAD. https://www.britishcouncil.org/sites/default/files/tne_classification_framework-final.pdf Kosmützky, A., & Putty, R. (2016). Transcending borders and traversing boundaries: A systematic review of the literature on transnational, offshore, cross-border, and borderless higher education. Journal of Studies in International Education, 20(1), 8–33. https://doi.org/10.1177/1028315315604719 Manatos, M. J., & Huisman, J. (2020). The use of the European Standards and Guidelines by national accreditation agencies and local review panels. Quality in Higher Education, 26(1), 48–65. https://doi.org/10.1080/13538322.2020.1728835 Marginson, S. (2014). University rankings and social science. European Journal of Education, 49(1), 45–59. https://doi.org/10.1111/ejed.12061 Musselin, C. (2018). New forms of competition in higher education. Socio-Economic Review, 16(3), 657–683. https://doi.org/10.1093/ser/mwy033 Organisation for Economic Co-operation and Development & UNESCO. (2005). Guidelines for quality provision in cross-border higher education. OECD Publishing. https://www.oecd.org/en/publications/2005/12/guidelines-for-quality-provision-in-cross-border-higher-education_g1gh67fe.html Seyfried, M., & Pohlenz, P. (2018). Assessing quality assurance in higher education: Quality managers' perceptions of effectiveness. European Journal of Higher Education, 8(3), 258–271. https://doi.org/10.1080/21568235.2018.1474777 Shams, F., & Huisman, J. (2016). The role of institutional dual embeddedness in the strategic local adaptation of international branch campuses: Evidence from Malaysia and Singapore. Studies in Higher Education, 41(6), 955–970. https://doi.org/10.1080/03075079.2014.966669 Snyder, H. (2019). Literature review as a research methodology: An overview and guidelines. Journal of Business Research, 104, 333–339. https://doi.org/10.1016/j.jbusres.2019.07.039 Tasopoulou, K., & Tsiotras, G. (2017). Benchmarking towards excellence in higher education. Benchmarking: An International Journal, 24(3), 617–634. https://doi.org/10.1108/BIJ-03-2016-0036 Torraco, R. J. (2016). Writing integrative literature reviews: Using the past and present to explore the future. Human Resource Development Review, 15(4), 404–428. https://doi.org/10.1177/1534484316671606 UNESCO. (2019). Global convention on the recognition of qualifications concerning higher education. UNESCO. https://www.unesco.org/en/higher-education/global-convention Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Wilkins, S. (2021). Two decades of international branch campus development, 2000–2020: A review. International Journal of Educational Management, 35(1), 311–326. https://doi.org/10.1108/IJEM-08-2020-0409 Wilkins, S., & Huisman, J. (2012). The international branch campus as transnational strategy in higher education. Higher Education, 64(5), 627–645. https://doi.org/10.1007/s10734-012-9516-5 Zapp, M., & Ramirez, F. O. (2019). Beyond internationalisation and isomorphism: The construction of a global higher education regime. Comparative Education, 55(4), 473–493. https://doi.org/10.1080/03050068.2019.1638103 Hashtags: #SDG17 #PartnershipsForTheGoals #SDG4 #QualityEducation #TransnationalEducation #QualityAssurance #CrossBorderPartnerships #Benchmarking #Accreditation #InternationalBranchCampuses #JointDegrees #U7YJournal

Author: Amina Hassan Affiliation: Swiss International University (SIU) ORCID ID: 0009-0005-7237-7565 Submitted 02 April 2026; Revised 06 June 2026; Accepted 11 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10003 Volume 3, December 2026, (SpecSDG10003) Abstract Marine ecosystems are changing faster than the observation systems designed to track them. This article presents an integrative review of digital technology interventions in marine ecosystem monitoring, synthesising peer-reviewed research on sensing platforms, molecular and acoustic methods, machine learning, and ocean data infrastructures published mainly between 2014 and 2026. The review addresses three questions: which classes of digital technology now support marine ecosystem monitoring and what ecological information each supplies; which recurring technical and institutional constraints limit their contribution to sustained, decision-relevant observation; and how these classes can be combined conceptually into an architecture linking sensing to management decisions. The synthesis shows that satellite remote sensing, autonomous platforms, environmental DNA, and passive acoustics have matured within largely separate research streams, each with distinctive observational strengths and each constrained by calibration demands, energy and sensor limits, interpretive uncertainty, or fragmented data practices. Machine learning relieves an analysis bottleneck created by high-volume sensing but introduces its own validation requirements. Building on these findings, the article proposes a four-layer sensing-to-decision architecture that connects observation, analytics, data integration, and governance through explicit feedback loops, using essential ocean variables as the prioritisation logic that binds the layers together. The framework offers monitoring programme designers and ocean governance bodies a structured basis for investment and evaluation decisions, and it identifies cost-effectiveness evidence, cross-domain interoperability, and institutional data governance as priorities for future research. Keywords: marine ecosystem monitoring, ocean observing systems, environmental DNA, passive acoustic monitoring, machine learning, digital twin, autonomous platforms 1. Introduction Societies depend on the ocean for food, climate regulation, and economic activity, and this dependence is intensifying at the same time as marine ecosystems come under growing anthropogenic pressure (Ryabinin et al., 2019). Warming, acidification, and deoxygenation are altering ocean biogeochemistry at scales that conventional observation struggles to detect (Chai et al., 2020). Decisions about fisheries, protected areas, pollution control, and climate adaptation therefore rest on the quality, coverage, and timeliness of ecological information about the sea. The United Nations Decade of Ocean Science for Sustainable Development (2021–2030) makes this link explicit by calling for a more complete and sustainable observing system whose outputs feed science-based decision making (Ryabinin et al., 2019). The established toolkit for marine ecological observation was not designed for this task. Classical monitoring based on research vessels, towed nets, grabs, and diver surveys offers low spatial and temporal resolution and is labour-intensive for every unit of area and time surveyed (Danovaro et al., 2016). A review of monitoring practice in the United Kingdom found that many techniques in routine use had changed little since the early 1900s (Bean et al., 2017). Ship-based programmes remain indispensable for calibration and for long-term comparability, but they cannot alone deliver the sustained, high-resolution, multi-variable coverage that ecosystem-based management requires. Over the past decade a set of digital technologies has matured to the point where it plausibly changes what monitoring programmes can observe. Satellite ocean colour instruments track phytoplankton dynamics at seasonal and interannual scales (Groom et al., 2019). Autonomous floats, gliders, and underwater vehicles carry physical, biogeochemical, and imaging sensors into regions ships rarely reach (Roemmich et al., 2019; Testor et al., 2019; Wynn et al., 2014). Environmental DNA (eDNA) metabarcoding detects communities of organisms from water samples without capturing a single animal (Miya, 2022). Passive acoustic monitoring converts the ocean soundscape into biodiversity information (Mooney et al., 2020). Deep learning methods process the resulting volumes of imagery and audio (Goodwin et al., 2022; Stowell, 2022), and data infrastructures and digital twins promise to fuse these streams into decision-ready products (Miedtank et al., 2024; Tanhua et al., 2019). The literature documenting these developments has a structural weakness: it is organised almost entirely within technology classes. Authoritative reviews exist for eDNA (Deiner et al., 2017), passive acoustics (Mooney et al., 2020), autonomous vehicles (Whitt et al., 2020; Wynn et al., 2014), satellite observation (Groom et al., 2019; Hedley et al., 2016), and deep learning in marine ecology (Goodwin et al., 2022), yet each stream defines its own problems, evaluates its own constraints, and rarely engages the others. The two broad cross-technology syntheses available are jurisdictionally bounded, one to European regulatory assessment (Danovaro et al., 2016) and one to the United Kingdom (Bean et al., 2017), and both predate the maturation of deep learning pipelines and marine digital twins. The literature on analytics and data governance, meanwhile, has developed separately from the literature on sensing platforms (Malde et al., 2020; Tanhua et al., 2019). What the field has not resolved is how the technology classes relate to one another as components of a single observation-to-decision chain: which class supplies which ecological variables, which constraints recur across classes, and what an integrated architecture would need to look like for monitoring outputs to reach management. This unresolved question matters directly to the designers of national and regional monitoring programmes, to coordinating bodies for ocean observation, and to funders who must allocate resources across platforms, analytics, and data systems rather than within any single technology. This article addresses that gap through an integrative review with a conceptual contribution. It synthesises verified peer-reviewed literature across four technology domains, sensing platforms, molecular and acoustic methods, machine learning analytics, and data integration and governance, and from that synthesis derives a layered sensing-to-decision architecture for digital marine ecosystem monitoring. The review answers three research questions: RQ1. Which classes of digital technology now support marine ecosystem monitoring, and what ecological information does each class supply? RQ2. Which recurring technical and institutional constraints limit the contribution of these technologies to sustained, decision-relevant monitoring? RQ3. How can these technology classes be combined conceptually into an integrated architecture that links sensing to management decisions? The remainder of the article proceeds as follows. Section 2 situates the review in the literature on sustained ocean observation and identifies the integration deficit in existing research. Section 3 sets out the integrative review protocol. Section 4 presents the findings in five subsections, including the synthesis table and the conceptual framework. Section 5 discusses theoretical, practical, and policy implications, states the limitations of the review, and outlines future research. Section 6 concludes. 2. Literature Review and Theoretical Framework 2.1 From expeditions to sustained ocean observation The intellectual context for digital monitoring technology is the long transition from expedition-based oceanography to sustained, systematic observation. The Argo programme of profiling floats is the emblem of this transition, and its designers now propose extending the array beyond its original 2000 m depth limit toward a full-depth, multidisciplinary global system with improved coverage of equatorial and boundary regions (Roemmich et al., 2019). The biogeochemical extension of Argo equips floats with sensors for six variables, oxygen, nitrate, pH, chlorophyll a, suspended particles, and downwelling irradiance, and releases the data freely within 24 hours of transmission (Claustre et al., 2020). Underwater gliders acquired a comparable coordinating structure when the OceanGliders programme, established in 2016, organised glider observation of physical, biogeochemical, and biological processes within the Global Ocean Observing System (Testor et al., 2019). Sustained observation requires agreement on what to measure. Miloslavich et al. (2018) derived a set of biological and ecosystem essential ocean variables (EOVs), including plankton biomass and diversity, the abundance and distribution of fish, turtles, birds, and mammals, and the cover of coral, seagrass, mangrove, and macroalgal habitats, selecting variables through a driver-pressure-state-impact-response logic weighted by impact and feasibility. The EOV concept supplies a shared prioritisation language across otherwise disparate observing communities, and it recurs throughout this review as the natural bridge between technology classes. At the policy level, the UN Decade frames the purpose of these systems as feeding a science-based decision-making apparatus rather than accumulating data for its own sake (Ryabinin et al., 2019). 2.2 Technology-specific research streams Within this broad movement, distinct research streams have developed around individual technology classes. A robotics and platforms stream traces the contribution of autonomous underwater vehicles to seafloor science (Wynn et al., 2014), articulates a collective vision for autonomous observation (Whitt et al., 2020), and designs permanent deep-sea networks that combine cabled seabed platforms with mobile vehicles (Aguzzi et al., 2019). A remote sensing stream evaluates satellite ocean colour as the principal synoptic view of marine ecosystems (Groom et al., 2019), reviews reef-scale applications (Hedley et al., 2016), and specifies the sensor characteristics that coastal biodiversity observation would require (Muller-Karger et al., 2018). A molecular stream established environmental DNA as a conservation monitoring tool (Thomsen & Willerslev, 2015), codified the metabarcoding workflow (Deiner et al., 2017), and demonstrated its application to marine fish communities (Miya, 2022). An acoustics stream assesses passive listening as a biodiversity method (Mooney et al., 2020) against the backdrop of a rapidly changing ocean soundscape (Duarte et al., 2021), with computational bioacoustics supplying the analytical machinery (Stowell, 2022). A machine learning stream diagnoses the widening gap between data acquisition and analytic capacity (Malde et al., 2020) and maps deep learning applications across marine ecology (Goodwin et al., 2022). Finally, a data stream applies the FAIR principles of findability, accessibility, interoperability, and reusability to ocean data services (Tanhua et al., 2019) and, most recently, explores marine digital twins (Miedtank et al., 2024) following broader proposals for digital twins of the Earth system (Bauer et al., 2021). Each stream is internally rigorous, and several include self-critical assessments of their own limits: acoustic researchers urge caution about current biodiversity indices (Mooney et al., 2020), molecular ecologists document how protocol choices propagate into community-level conclusions (Deiner et al., 2017), and remote sensing scientists detail the calibration burden behind apparently effortless global imagery (Groom et al., 2019). What the streams share is an outward-facing claim, namely that their technology serves ecosystem monitoring, combined with limited analysis of how that service depends on the other streams. 2.3 The integration deficit Two earlier syntheses attempted a cross-technology view. Danovaro et al. (2016) reviewed molecular methods, in situ instrumentation, and remote sensing against European marine-status assessment obligations and concluded that recently developed technologies can offer advantages in accuracy, efficiency, and cost over classical methods. Bean et al. (2017) catalogued platforms and sensors used in United Kingdom monitoring and argued that integrating traditional techniques with technological and modelling innovation could yield data that are currently too difficult or expensive to gather. Both reviews are valuable precedents, and both are explicitly bounded: their organising questions are set by a specific regulatory regime, and both were written before deep learning pipelines, biogeochemical float networks at global scale, and digital twin initiatives reshaped the field. The theoretical lens adopted here treats a monitoring system as a value chain in which raw signals become ecological variables, variables become synthesised knowledge, and knowledge becomes management action. This framing is implicit in the observing-system literature: the EOV framework connects measurement to societal requirements (Miloslavich et al., 2018), FAIR data services connect measurement to reuse (Tanhua et al., 2019), and the Decade connects observation to decision making (Ryabinin et al., 2019). Making the chain explicit, and locating each technology class within it, is the conceptual work this review undertakes. The review is therefore built to answer the gap identified above: a technology-by-technology synthesis oriented not toward any single class but toward the connective structure among classes. 3. Method 3.1 Review design The study follows the integrative review method, which is appropriate when a field's evidence is dispersed across heterogeneous literatures and the intended output is a new conceptual structure rather than a pooled effect estimate (Torraco, 2016; Whittemore & Knafl, 2005). Integrative reviews permit the combination of diverse source types, including empirical studies, methodological reviews, and programmatic position papers, under an explicit analytical framework (Whittemore & Knafl, 2005). Because the guiding questions span technical and institutional issues, an integrative design was preferred over a narrowly systematic one, consistent with guidance that the review form should match the research purpose (Snyder, 2019). 3.2 Search strategy Literature was identified through structured searches of Scopus and the Web of Science Core Collection, supplemented by Google Scholar for forward and backward citation chasing. Example search strings included: ("marine monitoring" OR "ocean observing" OR "marine ecosystem" ) AND ("autonomous underwater vehicle" OR glider OR "profiling float" OR "remote sensing" OR "ocean colour"); ("environmental DNA" OR metabarcoding) AND (marine OR ocean) AND (monitor* OR biodiversity); ("passive acoustic monitoring" OR soundscape OR bioacoustics) AND (marine OR ocean); ("machine learning" OR "deep learning") AND (marine OR ocean) AND (imagery OR acoustics OR ecology); and ("digital twin" OR "FAIR data" OR "essential ocean variables") AND ocean. The core search window was January 2014 to August 2026, reflecting the period in which the reviewed technologies reached operational maturity, with older works admitted when they were methodologically foundational for the review design or for a technology class. 3.3 Inclusion and exclusion criteria Sources were included when they (a) were peer-reviewed journal publications in English; (b) addressed digital technologies for observing marine ecosystems, or the analytic and data infrastructures that such observation requires; and (c) operated at a synthesis, roadmap, or programme level capable of informing cross-technology comparison, rather than reporting a single-site engineering trial. Sources were excluded when they concerned exclusively freshwater systems, addressed digital twins of ships or industrial installations rather than of the marine environment, treated algorithms with no marine application, or could not be verified against the Crossref bibliographic record. Every retained source had its full bibliographic metadata confirmed against the Crossref application programming interface, and sources whose records could not be confirmed were discarded. Screening proceeded in two stages, first on title and abstract and then on full available text, against the criteria above. The final corpus comprises 30 sources: 27 substantive works on marine monitoring technology and infrastructure and three methodological anchors for the review design. 3.4 Analysis and framework derivation Retained sources were coded into an extraction matrix with fields for technology class, observed ecological properties, reported strengths, reported constraints, data practices, and stated links to management or governance. Synthesis followed the constant comparison logic recommended for integrative reviews, iterating between within-class summaries and cross-class comparison until stable themes emerged (Whittemore & Knafl, 2005). Four themes were retained: platform-based sensing, molecular and acoustic sensing, machine learning analytics, and data integration and governance. The conceptual framework was derived abductively: the layered structure was proposed as the simplest arrangement that accounts for the relationships documented in the matrix, then tested against each source for contradiction and refined. Table 1 and Figure 1 are direct products of this process. 3.5 Rigor and trustworthiness Several measures bound the trustworthiness of the review. All bibliographic records were verified against Crossref before citation, and an audit trail links every substantive claim to a logged source. The review does not claim exhaustiveness: integrative synthesis prioritises conceptual coverage over census-style completeness, and no article-count flow diagram is reported because the corpus was consolidated for conceptual saturation rather than enumerated through a preregistered protocol. Restricting the corpus to English-language, journal-published sources narrows the evidence base, and reliance on synthesis-level papers risks importing the optimism characteristic of community position statements; the findings sections counter this by giving constraints equal standing with capabilities. Screening and coding were performed by a single author, a limitation addressed through explicit criteria and the documented audit trail rather than through inter-coder statistics. 4. Findings The four thematic subsections below answer RQ1 and RQ2 by technology class, and Section 4.5 answers RQ3. Table 1 summarises the synthesis: for each class it lists what is observed, the principal strengths reported in the reviewed literature, and the constraints that recur across sources. Table 1 Digital Technology Classes for Marine Ecosystem Monitoring: Observed Properties, Strengths, Constraints, and Sources Technology class Primary observed properties Principal strengths Recurring constraints Key sources Satellite remote sensing Ocean colour and chlorophyll; phytoplankton dynamics; reef extent, benthic cover, sea surface temperature Synoptic, repeated coverage; operational monitoring of blooms and eutrophication Mission continuity and cross-sensor merging; stringent calibration; coastal resolution gaps Groom et al. (2019); Hedley et al. (2016); Muller-Karger et al. (2018) Profiling floats and gliders Temperature and salinity; oxygen, nitrate, pH, chlorophyll a, particles, irradiance Sustained interior observation; open data within 24 hours; float array building toward global coverage Biogeochemical sensors at lower technology readiness; glider deployments largely regional and near-surface Roemmich et al. (2019); Claustre et al. (2020); Testor et al. (2019); Chai et al. (2020) AUVs and seafloor observatories Seafloor imagery and habitat maps; continuous deep-sea video and acoustic observation High-resolution mapping of environments inaccessible to surface vessels; continuous presence Subsurface energy limits; biofouling; acoustic communication bandwidth; cost and expertise Wynn et al. (2014); Aguzzi et al. (2019); Whitt et al. (2020) Environmental DNA metabarcoding Multi-species community composition; marine fish richness and spatiotemporal dynamics Noninvasive; cost-effective; multi-taxon detection amid declining taxonomic expertise Relative abundance estimation unresolved; sensitivity to workflow choices from primers to bioinformatics Thomsen & Willerslev (2015); Deiner et al. (2017); Miya (2022) Passive acoustic monitoring Vocalising species; anthropogenic noise; ocean soundscapes Long-duration listening in a medium where sound travels far Current biodiversity indices require caution; large archives depend on machine learning analysis Mooney et al. (2020); Duarte et al. (2021); Stowell (2022) Machine learning analytics Detection, classification, tracking, segmentation of organisms in imagery and audio Relieves the analysis bottleneck created by high-volume sensing Dependence on supervised training data; standardisation and validation across programmes Malde et al. (2020); Goodwin et al. (2022); Lombard et al. (2019) Data infrastructure and digital twins Integrated essential-ocean-variable products; FAIR data services; marine digital twins Reuse and comparability of observations; route from observation to decision support Fragmented catalogues; uneven maturity across disciplines; sustained funding for data management Tanhua et al. (2019); Miloslavich et al. (2018); Miedtank et al. (2024) Note. Synthesis by the author from the sources cited in each row. Strengths and constraints are those reported in the cited literature; the table introduces no new empirical data. AUV = autonomous underwater vehicle; FAIR = findable, accessible, interoperable, reusable. 4.1 Platform-based sensing from orbit to seafloor Satellite remote sensing supplies the only routinely synoptic view of marine ecosystems. Ocean colour radiometry underpins the study of phytoplankton dynamics at seasonal and interannual scales and supports operational monitoring of coastal eutrophication, harmful algal blooms, and sediment plumes (Groom et al., 2019). For structured habitats, satellite and airborne observation maps reef extent, benthic cover, and sea surface temperature, trading the point-level accuracy of field survey for complete spatial coverage (Hedley et al., 2016). The constraints of the class are equally well documented. The long-term ocean colour record is stitched together from a sequence of one-off missions whose merging risks artefacts, and retrieval is demanding because the water-leaving signal constitutes less than a tenth of what the sensor detects at the top of the atmosphere (Groom et al., 2019). For coastal biodiversity specifically, Muller-Karger et al. (2018) concluded that current and planned satellites are not designed to observe rapidly changing coastal essential biodiversity variables, and specified requirements, including pixel sizes of 30 to 100 m and revisit times of hours to days, that no operational system yet meets. In the water column, autonomous platforms have converted what was once expedition sampling into sustained observation. The proposed extension of Argo toward a full-depth, biogeochemically instrumented global array (Roemmich et al., 2019), and the demonstrated capacity of biogeochemical floats to deliver six-variable profiles as open data within a day of collection (Claustre et al., 2020), define the current benchmark for scalable in situ observation. Reviews of this platform ecosystem argue that autonomous mobile assets combined with satellite data can provide the four-dimensional information needed to model and forecast ocean health under warming, acidification, and deoxygenation (Chai et al., 2020). Gliders contribute sustained sections through boundary currents and storm-affected regions under international coordination (Testor et al., 2019), while autonomous underwater vehicles deliver seafloor imaging and habitat mapping at resolutions unattainable from surface vessels, including under ice and in the deep sea (Wynn et al., 2014). At the seafloor itself, proposed monitoring networks combine fixed cabled platforms with mobile vehicles carrying video, acoustic, and molecular payloads for continuous ecological observation (Aguzzi et al., 2019), aligned with proposals for globally coordinated deep-ocean monitoring variables (Danovaro et al., 2020). The recurring constraints of the platform class are physical and economic. Subsurface energy availability fundamentally limits propulsion, sensing, and communication; biological and biogeochemical sensors mostly sit at lower technology readiness than their physical counterparts; biofouling degrades long deployments; underwater acoustic communication restricts data return; and acquisition and operation remain costly and expertise-intensive (Whitt et al., 2020). These constraints explain a pattern visible across the corpus: while the float network builds toward global biogeochemical coverage, glider and surface-vehicle deployments carrying such sensors have largely remained regional and near-surface (Chai et al., 2020). 4.2 Molecular and acoustic sensing Environmental DNA and passive acoustics extend monitoring to organisms and behaviours that platforms with cameras and physical sensors cannot efficiently capture. eDNA, defined as genetic material obtained directly from environmental samples, offers a noninvasive and standardisable survey approach at a time when taxonomic expertise for morphological identification is in decline (Thomsen & Willerslev, 2015). Metabarcoding of such samples surveys whole communities of animals and plants and has been applied across marine, freshwater, and terrestrial systems (Deiner et al., 2017). For marine fish specifically, eDNA metabarcoding detects multiple species simultaneously and resolves species richness and spatiotemporal community dynamics across spatial scales, at costs the review literature characterises as low relative to conventional survey (Miya, 2022). The interpretive constraints are handled candidly in the same literature: estimating relative abundance from read counts remains problematic, and results are sensitive to choices along the entire workflow from primers and library preparation to bioinformatic filtering (Deiner et al., 2017). Integration with platforms is already envisaged, with in situ sequencing instruments included in deep-sea observatory designs to detect organisms, including prokaryotes, that imaging cannot record (Aguzzi et al., 2019). Passive acoustic monitoring exploits the physics of the ocean, where sound travels faster and farther than in air and where marine organisms have evolved to rely on acoustic communication (Duarte et al., 2021). Passive recorders therefore capture vocalising species together with the anthropogenic sound, whose growth Duarte et al. (2021) document as a defining feature of the modern ocean soundscape. As a biodiversity method, however, the field's own assessment is cautious: Mooney et al. (2020) recommend restraint in applying current acoustic diversity indices while identifying machine learning and source separation as the most promising analytical directions. This caution matters for programme design, because it separates the mature use of acoustics for detecting target species and anthropogenic noise from the still-developing use of acoustics as a general biodiversity metric. 4.3 Machine learning and the analysis bottleneck The sensing classes reviewed above share one systemic consequence: they generate data faster than conventional analysis can absorb. Malde et al. (2020) diagnose this directly, observing that modern sensor systems produce unprecedented data volumes while analytic capacity has not kept pace, and positioning deep learning as the most plausible response across marine science. The applications literature substantiates the response. Deep networks now perform detection, classification, tracking, and segmentation on marine imagery of plankton, fish, and marine mammals (Goodwin et al., 2022). In bioacoustics, deep learning inherited its methods from speech and image processing, and the field faces problems distinct enough from those parent domains that Stowell (2022) proposes a dedicated research roadmap for computational bioacoustics. In plankton observation, imaging instruments and optical sensors are mature, but inconsistent and diverse methods across programmes obstruct the detection of global and long-term patterns, prompting calls for standardised, quality-controlled, automated pipelines integrated with existing observing infrastructure (Lombard et al., 2019). Read across sources, the machine learning theme carries a double message. Analytics is the multiplier that makes high-volume sensing worthwhile; without it, imagery and audio accumulate as unprocessed archives. At the same time, the reviewed literature ties the value of these methods to conditions that are institutional as much as algorithmic: the applications surveyed rely largely on supervised learning, which binds performance to curated training data (Goodwin et al., 2022), and outputs only become comparable across programmes where processing and quality control are standardised (Lombard et al., 2019). The bottleneck, in short, moves rather than disappears: from analysing data to governing the analysis. 4.4 Data integration, digital twins, and governance The final theme concerns the infrastructure through which observations become usable evidence. The FAIR framework, requiring ocean data to be findable, accessible, interoperable, and reusable, is the organising standard, and its application reveals uneven maturity: physical oceanography has advanced furthest, aided by the Argo data system's automated quality control and global assembly centres, while biological and biogeochemical data communities lag, and infrastructure remains fragmented, illustrated by the existence of more than 70 data catalogues for polar regions alone (Tanhua et al., 2019). The same source proposes a concrete governance norm, recommending that 5 to 10 percent of research funding be dedicated to data management (Tanhua et al., 2019). The EOV framework complements FAIR by specifying what merits sustained observation in the first place, derived from societal drivers and feasibility rather than disciplinary habit (Miloslavich et al., 2018). Digital twins represent the most ambitious integration proposal in the corpus. Following the argument for a digital twin of Earth in support of the green transition (Bauer et al., 2021), marine digital twins are now advanced as a means of deepening ocean understanding (Miedtank et al., 2024). Within the sensing-to-decision chain, the twin concept occupies the apex: it presupposes interoperable data flows, validated analytics, and sustained observation, and its credibility therefore depends on every layer beneath it. The policy architecture for such integration exists in outline: the UN Decade explicitly aims to connect a more complete observing system to science-based decision making (Ryabinin et al., 2019), and national reviews argue that integrated deployment of old and new methods can satisfy legislative monitoring obligations at acceptable cost (Bean et al., 2017; Danovaro et al., 2016). 4.5 An integrated sensing-to-decision architecture The synthesis supports a conceptual answer to RQ3, presented as Figure 1. The framework is interpretive: it is derived from the reviewed literature but is itself a proposal, not an empirical result. It arranges digital marine monitoring into four layers connected by two feedback loops, with two cross-cutting conditions. Figure 1. A sensing-to-decision architecture for digital marine ecosystem monitoring. Solid arrows show the flow of data and derived products; dashed arrows show the prioritisation and validation feedback loops; the dashed bar lists cross-cutting enabling conditions.Original figure by the author, synthesised from the reviewed literature. The observation layer comprises the complementary sensing classes established in Sections 4.1 and 4.2: spaceborne remote sensing for synoptic coverage, autonomous and fixed platforms for interior and seafloor observation, and molecular and acoustic methods for biological detail. The analytics layer converts raw signals into ecological variables through machine learning pipelines whose outputs are only as credible as their training data, standardisation, and validation (Goodwin et al., 2022; Lombard et al., 2019). The integration layer assembles validated variables into FAIR data services (Tanhua et al., 2019) and, at its most ambitious, into marine digital twins (Miedtank et al., 2024). The decision layer uses these products in assessment, management, and policy, the destination the Decade assigns to ocean observation (Ryabinin et al., 2019). Two feedback loops make the architecture a system rather than a pipeline. A prioritisation loop runs from the decision layer back to observation: essential ocean variables, derived from societal requirements, determine what the observation layer should measure and which sensor investments matter most (Miloslavich et al., 2018). A validation loop runs between layers: legacy ship-based and station-based methods calibrate new sensors and anchor long-term comparability, a role the national syntheses identify as the enduring value of traditional monitoring (Bean et al., 2017; Danovaro et al., 2016). The cross-cutting conditions are sustained funding with an explicit data-management share (Tanhua et al., 2019) and capacity development so that participation in digitally intensive monitoring is not confined to wealthy programmes (Ryabinin et al., 2019; Tanhua et al., 2019). The architecture answers RQ3 by specifying where each technology class sits, what it depends on, and where investment imbalances break the chain: sensing without analytics accumulates archives, analytics without data services produces incomparable outputs, and integration without decision-layer uptake produces infrastructure without consequence. 5. Discussion 5.1 Theoretical implications The review's first implication is that marine monitoring technology is best theorised as a coupled system rather than a set of substitutable instruments. The technology-class literatures reviewed here each present their own tools as monitoring solutions, yet the constraints they report are almost always resolved in a different layer of the architecture: the platform literature's data-volume problem is addressed by analytics (Malde et al., 2020), the analytics literature's standardisation problem is addressed by data governance (Lombard et al., 2019; Tanhua et al., 2019), and the governance literature's prioritisation problem is addressed by variable frameworks negotiated with decision makers (Miloslavich et al., 2018). Cross-layer dependence, not within-class capability, is the analytically productive unit. Second, the EOV framework functions in this synthesis as a boundary object: it is the one construct that every layer can address, from sensor design to data services to policy reporting (Miloslavich et al., 2018). Treating EOVs as the binding logic of the architecture extends their original role in observing-system coordination into a general integration mechanism for digital monitoring. Third, the framework extends the earlier cross-technology syntheses beyond their jurisdictional origins. Danovaro et al. (2016) and Bean et al. (2017) demonstrated integration arguments within specific regulatory settings; the architecture proposed here generalises the integration claim into a structure that is independent of any single regime and explicitly incorporates the analytics and integration layers that postdate those reviews. 5.2 Practical and policy implications For monitoring programme designers, the architecture converts a diffuse technology menu into a portfolio question: investments should be balanced across layers, because the marginal value of an additional sensor depends on the analytics, data services, and decision pathways available to absorb its output. The reviewed literature supplies concrete anchors for such portfolios: open release of float data within a day of collection as a benchmark for data latency (Claustre et al., 2020), a 5 to 10 percent share of research funding for data management (Tanhua et al., 2019), and continued operation of established methods alongside digital ones to preserve the comparability of long-term series (Bean et al., 2017). For ocean governance bodies, two implications stand out. First, the caution voiced within the acoustic and molecular literatures (Deiner et al., 2017; Mooney et al., 2020) argues for standards that distinguish decision-grade applications, such as target-species detection, from research-grade applications, such as community-level indices still under validation. Second, the Decade's window for coordinated investment (Ryabinin et al., 2019) makes the capacity dimension urgent: without deliberate transfer of tools, training data, and infrastructure, digitally intensive monitoring risks concentrating observational capability in already well-resourced regions, a concern consistent with the capacity-building emphasis in the ocean data literature (Tanhua et al., 2019). 5.3 Limitations The limitations of this review follow from its method. First, an integrative review is selective by design; although every source was verified and the corpus was assembled to cover the field's principal streams, other defensible corpora exist, and a different selection could weight the themes differently (Snyder, 2019; Whittemore & Knafl, 2005). Second, the corpus favours synthesis-level and programmatic publications, which document capabilities and roadmaps more readily than operational failures; the balance of strengths and constraints reported here is therefore only as honest as the self-assessment of the reviewed communities. Third, restriction to English-language journal literature excludes grey literature from monitoring agencies and non-English national programmes in which considerable operational experience resides. Fourth, screening and coding were conducted by a single author, and no inter-coder reliability can be reported. Finally, the field moves quickly; conclusions about maturity, particularly for machine learning and digital twins, carry short half-lives. 5.4 Future research Four directions follow directly from what the reviewed literature leaves open. First, comparative cost-effectiveness: the corpus contains claims of efficiency advantages for new methods (Danovaro et al., 2016; Miya, 2022) but no cross-class economic evidence base adequate for portfolio decisions; empirical cost-per-variable studies across technology classes are needed. Second, interoperability of molecular and acoustic outputs with the EOV framework: community-level eDNA and acoustic indices require the validation programmes that their own literatures call for before they can enter sustained observing systems (Deiner et al., 2017; Mooney et al., 2020). Third, the transferability of machine learning models across regions, instruments, and time, identified as a live problem in both marine imagery and bioacoustics (Goodwin et al., 2022; Stowell, 2022), warrants systematic evaluation under operational monitoring conditions. Fourth, the institutional side of the architecture, including who sustains data services, how the recommended budget shares fare in practice, and whether digital twins alter decisions rather than dashboards, is essentially unstudied in the reviewed corpus and requires governance-oriented empirical research. Against the gap stated in the Introduction, the review closes the conceptual portion: it maps the technology classes onto a common architecture, identifies the cross-class constraints, and supplies the integration structure that the class-specific literatures lacked. The empirical portion of the gap, above all the comparative economics and the institutional performance of integrated systems, remains open and defines the agenda above. 6. Conclusion This integrative review synthesised the peer-reviewed literature on digital technology in marine ecosystem monitoring across four domains: sensing platforms from satellites to seafloor observatories, molecular and acoustic methods, machine learning analytics, and data integration and governance. The synthesis shows a field of individually maturing technology classes whose monitoring value is jointly determined: each class resolves its central constraint only through the layers around it. The proposed sensing-to-decision architecture organises these dependencies into four layers bound by prioritisation and validation feedback loops, with essential ocean variables as the connective logic and sustained funding and capacity development as enabling conditions. The framework gives programme designers a structured basis for balancing investment across sensing, analytics, and data infrastructure, and gives researchers a map of the seams between literatures where the most consequential open questions, from cross-class cost-effectiveness to the institutional performance of digital twins, are located. Marine monitoring is becoming a digital system; understanding it, and funding it, as a system is the condition for that transition to serve ocean governance. Declarations Funding: This research received no external funding. Conflicts of Interest: The author declares no conflict of interest. Ethics: This study is a review of published literature and did not involve human participants, animal subjects, or primary data collection; ethical approval was therefore not required. Data Availability: No new data were created or analysed in this study. All sources synthesised are cited in the reference list and publicly identifiable through the digital object identifiers provided. References Aguzzi, J., Chatzievangelou, D., Marini, S., Fanelli, E., Danovaro, R., Flöegel, S., Lebris, N., Juanes, F., De Leo, F. C., Del Rio, J., Thomsen, L., Costa, C., Riccobene, G., Tamburini, C., Lefevre, D., Gojak, C., Poulain, P.-M., Favali, P., Griffa, A., … Company, J. B. (2019). New high-tech flexible networks for the monitoring of deep-sea ecosystems. Environmental Science & Technology, 53(12), 6616–6631. https://doi.org/10.1021/acs.est.9b00409 Bauer, P., Stevens, B., & Hazeleger, W. (2021). A digital twin of Earth for the green transition. Nature Climate Change, 11(2), 80–83. https://doi.org/10.1038/s41558-021-00986-y Bean, T. P., Greenwood, N., Beckett, R., Biermann, L., Bignell, J. P., Brant, J. L., Copp, G. H., Devlin, M. J., Dye, S., Feist, S. W., Fernand, L., Foden, D., Hyder, K., Jenkins, C. M., van der Kooij, J., Kröger, S., Kupschus, S., Leech, C., Leonard, K. S., … Righton, D. (2017). A review of the tools used for marine monitoring in the UK: Combining historic and contemporary methods with modeling and socioeconomics to fulfill legislative needs and scientific ambitions. Frontiers in Marine Science, 4, 263. https://doi.org/10.3389/fmars.2017.00263 Chai, F., Johnson, K. S., Claustre, H., Xing, X., Wang, Y., Boss, E., Riser, S., Fennel, K., Schofield, O., & Sutton, A. (2020). Monitoring ocean biogeochemistry with autonomous platforms. Nature Reviews Earth & Environment, 1(6), 315–326. https://doi.org/10.1038/s43017-020-0053-y Claustre, H., Johnson, K. S., & Takeshita, Y. (2020). Observing the global ocean with Biogeochemical-Argo. Annual Review of Marine Science, 12, 23–48. https://doi.org/10.1146/annurev-marine-010419-010956 Danovaro, R., Carugati, L., Berzano, M., Cahill, A. E., Carvalho, S., Chenuil, A., Corinaldesi, C., Cristina, S., David, R., Dell'Anno, A., Dzhembekova, N., Garcés, E., Gasol, J. M., Goela, P., Féral, J.-P., Ferrera, I., Forster, R. M., Kurekin, A. A., Rastelli, E., … Borja, A. (2016). Implementing and innovating marine monitoring approaches for assessing marine environmental status. Frontiers in Marine Science, 3, 213. https://doi.org/10.3389/fmars.2016.00213 Danovaro, R., Fanelli, E., Aguzzi, J., Billett, D., Carugati, L., Corinaldesi, C., Dell'Anno, A., Gjerde, K., Jamieson, A. J., Kark, S., McClain, C., Levin, L., Levin, N., Ramirez-Llodra, E., Ruhl, H., Smith, C. R., Snelgrove, P. V. R., Thomsen, L., Van Dover, C. L., & Yasuhara, M. (2020). Ecological variables for developing a global deep-ocean monitoring and conservation strategy. Nature Ecology & Evolution, 4(2), 181–192. https://doi.org/10.1038/s41559-019-1091-z Deiner, K., Bik, H. M., Mächler, E., Seymour, M., Lacoursière-Roussel, A., Altermatt, F., Creer, S., Bista, I., Lodge, D. M., de Vere, N., Pfrender, M. E., & Bernatchez, L. (2017). Environmental DNA metabarcoding: Transforming how we survey animal and plant communities. Molecular Ecology, 26(21), 5872–5895. https://doi.org/10.1111/mec.14350 Duarte, C. M., Chapuis, L., Collin, S. P., Costa, D. P., Devassy, R. P., Eguiluz, V. M., Erbe, C., Gordon, T. A. C., Halpern, B. S., Harding, H. R., Havlik, M. N., Meekan, M., Merchant, N. D., Miksis-Olds, J. L., Parsons, M., Predragovic, M., Radford, A. N., Radford, C. A., Simpson, S. D., … Juanes, F. (2021). The soundscape of the Anthropocene ocean. Science, 371(6529), eaba4658. https://doi.org/10.1126/science.aba4658 Goodwin, M., Halvorsen, K. T., Jiao, L., Knausgård, K. M., Martin, A. H., Moyano, M., Oomen, R. A., Rasmussen, J. H., Sørdalen, T. K., & Thorbjørnsen, S. H. (2022). Unlocking the potential of deep learning for marine ecology: Overview, applications, and outlook. ICES Journal of Marine Science, 79(2), 319–336. https://doi.org/10.1093/icesjms/fsab255 Groom, S., Sathyendranath, S., Ban, Y., Bernard, S., Brewin, R., Brotas, V., Brockmann, C., Chauhan, P., Choi, J., Chuprin, A., Ciavatta, S., Cipollini, P., Donlon, C., Franz, B., He, X., Hirata, T., Jackson, T., Kampel, M., Krasemann, H., … Wang, M. (2019). Satellite ocean colour: Current status and future perspective. Frontiers in Marine Science, 6, 485. https://doi.org/10.3389/fmars.2019.00485 Hedley, J., Roelfsema, C., Chollett, I., Harborne, A., Heron, S., Weeks, S., Skirving, W., Strong, A., Eakin, C., Christensen, T., Ticzon, V., Bejarano, S., & Mumby, P. (2016). Remote sensing of coral reefs for monitoring and management: A review. Remote Sensing, 8(2), 118. https://doi.org/10.3390/rs8020118 Lombard, F., Boss, E., Waite, A. M., Vogt, M., Uitz, J., Stemmann, L., Sosik, H. M., Schulz, J., Romagnan, J.-B., Picheral, M., Pearlman, J., Ohman, M. D., Niehoff, B., Möller, K. O., Miloslavich, P., Lara-López, A., Kudela, R., Lopes, R. M., Kiko, R., … Appeltans, W. (2019). Globally consistent quantitative observations of planktonic ecosystems. Frontiers in Marine Science, 6, 196. https://doi.org/10.3389/fmars.2019.00196 Malde, K., Handegard, N. O., Eikvil, L., & Salberg, A.-B. (2020). Machine intelligence and the data-driven future of marine science. ICES Journal of Marine Science, 77(4), 1274–1285. https://doi.org/10.1093/icesjms/fsz057 Miedtank, A., Schneider, J., Manss, C., & Zielinski, O. (2024). Marine digital twins for enhanced ocean understanding. Remote Sensing Applications: Society and Environment, 36, 101268. https://doi.org/10.1016/j.rsase.2024.101268 Miloslavich, P., Bax, N. J., Simmons, S. E., Klein, E., Appeltans, W., Aburto-Oropeza, O., Andersen Garcia, M., Batten, S. D., Benedetti-Cecchi, L., Checkley, D. M., Jr., Chiba, S., Duffy, J. E., Dunn, D. C., Fischer, A., Gunn, J., Kudela, R., Marsac, F., Muller-Karger, F. E., Obura, D., & Shin, Y.-J. (2018). Essential ocean variables for global sustained observations of biodiversity and ecosystem changes. Global Change Biology, 24(6), 2416–2433. https://doi.org/10.1111/gcb.14108 Miya, M. (2022). Environmental DNA metabarcoding: A novel method for biodiversity monitoring of marine fish communities. Annual Review of Marine Science, 14, 161–185. https://doi.org/10.1146/annurev-marine-041421-082251 Mooney, T. A., Di Iorio, L., Lammers, M., Lin, T.-H., Nedelec, S. L., Parsons, M., Radford, C., Urban, E., & Stanley, J. (2020). Listening forward: Approaching marine biodiversity assessments using acoustic methods. Royal Society Open Science, 7(8), 201287. https://doi.org/10.1098/rsos.201287 Muller-Karger, F. E., Hestir, E., Ade, C., Turpie, K., Roberts, D. A., Siegel, D., Miller, R. J., Humm, D., Izenberg, N., Keller, M., Morgan, F., Frouin, R., Dekker, A. G., Gardner, R., Goodman, J., Schaeffer, B., Franz, B. A., Pahlevan, N., Mannino, A. G., … Jetz, W. (2018). Satellite sensor requirements for monitoring essential biodiversity variables of coastal ecosystems. Ecological Applications, 28(3), 749–760. https://doi.org/10.1002/eap.1682 Roemmich, D., Alford, M. H., Claustre, H., Johnson, K., King, B., Moum, J., Oke, P., Owens, W. B., Pouliquen, S., Purkey, S., Scanderbeg, M., Suga, T., Wijffels, S., Zilberman, N., Bakker, D., Baringer, M., Belbeoch, M., Bittig, H. C., Boss, E., … Yasuda, I. (2019). On the future of Argo: A global, full-depth, multi-disciplinary array. Frontiers in Marine Science, 6, 439. https://doi.org/10.3389/fmars.2019.00439 Ryabinin, V., Barbière, J., Haugan, P., Kullenberg, G., Smith, N., McLean, C., Troisi, A., Fischer, A., Aricò, S., Aarup, T., Pissierssens, P., Visbeck, M., Enevoldsen, H. O., & Rigaud, J. (2019). The UN Decade of Ocean Science for Sustainable Development. Frontiers in Marine Science, 6, 470. https://doi.org/10.3389/fmars.2019.00470 Snyder, H. (2019). Literature review as a research methodology: An overview and guidelines. Journal of Business Research, 104, 333–339. https://doi.org/10.1016/j.jbusres.2019.07.039 Stowell, D. (2022). Computational bioacoustics with deep learning: A review and roadmap. PeerJ, 10, e13152. https://doi.org/10.7717/peerj.13152 Tanhua, T., Pouliquen, S., Hausman, J., O'Brien, K., Bricher, P., de Bruin, T., Buck, J. J. H., Burger, E. F., Carval, T., Casey, K. S., Diggs, S., Giorgetti, A., Glaves, H., Harscoat, V., Kinkade, D., Muelbert, J. H., Novellino, A., Pfeil, B., Pulsifer, P. L., … Zhao, Z. (2019). Ocean FAIR data services. Frontiers in Marine Science, 6, 440. https://doi.org/10.3389/fmars.2019.00440 Testor, P., de Young, B., Rudnick, D. L., Glenn, S., Hayes, D., Lee, C. M., Pattiaratchi, C., Hill, K., Heslop, E., Turpin, V., Alenius, P., Barrera, C., Barth, J. A., Beaird, N., Bécu, G., Bosse, A., Bourrin, F., Brearley, J. A., Chao, Y., … Wilson, D. (2019). OceanGliders: A component of the integrated GOOS. Frontiers in Marine Science, 6, 422. https://doi.org/10.3389/fmars.2019.00422 Thomsen, P. F., & Willerslev, E. (2015). Environmental DNA: An emerging tool in conservation for monitoring past and present biodiversity. Biological Conservation, 183, 4–18. https://doi.org/10.1016/j.biocon.2014.11.019 Torraco, R. J. (2016). Writing integrative literature reviews: Using the past and present to explore the future. Human Resource Development Review, 15(4), 404–428. https://doi.org/10.1177/1534484316671606 Whitt, C., Pearlman, J., Polagye, B., Caimi, F., Muller-Karger, F., Copping, A., Spence, H., Madhusudhana, S., Kirkwood, W., Grosjean, L., Fiaz, B. M., Singh, S., Singh, S., Manalang, D., Sen Gupta, A., Maguer, A., Buck, J. J. H., Marouchos, A., Atmanand, M. A., … Khalsa, S. J. (2020). Future vision for autonomous ocean observations. Frontiers in Marine Science, 7, 697. https://doi.org/10.3389/fmars.2020.00697 Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Wynn, R. B., Huvenne, V. A. I., Le Bas, T. P., Murton, B. J., Connelly, D. P., Bett, B. J., Ruhl, H. A., Morris, K. J., Peakall, J., Parsons, D. R., Sumner, E. J., Darby, S. E., Dorrell, R. M., & Hunt, J. E. (2014). Autonomous underwater vehicles (AUVs): Their past, present and future contributions to the advancement of marine geoscience. Marine Geology, 352, 451–468. https://doi.org/10.1016/j.margeo.2014.03.012 Hashtags: #SDG14 #LifeBelowWater #SDG13 #ClimateAction #MarineEcosystemMonitoring #OceanObservingSystems #EnvironmentalDNA #PassiveAcousticMonitoring #MachineLearning #DigitalTwin #AutonomousPlatforms #U7YJournal #U7Y #AcademicResearch

Author: Noah Müller Affiliation: Swiss International University (SIU) ORCID ID: 0009-0002-7578-923X Submitted 17 April 2026; Revised 21 June 2026; Accepted 22 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10004 Volume 3, December 2026, (SpecSDG10004) Abstract Universities increasingly promote their sustainability commitments to prospective students through attention-seeking formats such as guerrilla interventions, viral social media campaigns, experiential events, and cause-led advocacy. Whether these disruptive promotional strategies are effective, and under what conditions they backfire, has not been systematically assessed. This article evaluates their efficacy through an integrative review that synthesises published evidence from three literatures that have developed largely in isolation: higher education marketing and branding, sustainability implementation and reporting in higher education, and green and disruptive marketing research. No new empirical data were collected. The synthesis indicates that documented efficacy is concentrated at the level of attention, engagement, and word of mouth, with the strongest higher-education evidence attached to interactive social media use; direct evidence linking disruptive sustainability promotion to enrolment remains absent. Efficacy is conditional rather than inherent: experimental green-marketing research shows that promotional claims outrunning verifiable performance depress brand attitudes and trust, and that third-party verification disciplines audience inferences. Because sustainability reporting and impact measurement in higher education remain uneven, many institutions currently lack the substantiation infrastructure that credible disruptive promotion requires. The article contributes a typology of disruptive promotional strategies with their documented outcomes, and a conceptual framework in which credibility conditions mediate the path from promotional format to student-facing outcomes while greenwashing perceptions threaten reversal. Implications for university marketers, institutional leaders, and researchers are developed, including an agenda for higher-education-specific causal evidence. Keywords: sustainability communication, higher education marketing, guerrilla marketing, greenwashing, university branding, student recruitment, integrative review 1. Introduction Two developments have converged on the university marketing office. First, higher education has consolidated into a competitive service market in which institutions actively manage brands, recruitment funnels, and reputational signals, even though scholarship on how universities actually market themselves has lagged behind practice (Chapleo, 2010; Hemsley-Brown & Oplatka, 2006). Second, sustainability has moved from the periphery of campus operations to the centre of institutional missions, curricula, and accountability frameworks, propelled by international policy agendas that assign education a formative role in sustainable development (Findler et al., 2019; Lozano et al., 2015; UNESCO, 2020). At the intersection of these developments sits a practical question with theoretical depth: whether disruptive promotional strategies can effectively market universities' sustainable practices to prospective students, and at what cost when they fail. The question is timely because the target audience is demonstrably hard to reach through conventional advertising. Research on adolescent users of social networking sites shows that advertising is actively avoided when it is perceived as irrelevant, when expectations are unmet, or when users are sceptical of the message or the medium (Kelly et al., 2010). Marketers in many sectors have responded with formats designed to breach indifference: guerrilla and ambient interventions built on surprise (Hutter & Hoffmann, 2014), shareable campaign content engineered for virality (Akpinar & Berger, 2017), experiential events (Zarantonello & Schmitt, 2013), and cause-led advocacy communication. Universities have adopted elements of this repertoire, most visibly on social media platforms (Bélanger et al., 2014; Peruta & Shields, 2017). Yet sustainability is a message category with distinctive failure modes. Experimental research on green advertising shows that promotional claims can depress brand attitudes when audiences doubt the claimant's underlying performance (Nyilasy et al., 2014), and that false environmental claims raise perceived greenwashing and damage evaluations (Schmuck et al., 2018). A tactic whose defining property is to amplify attention may therefore amplify scrutiny as well. The research gap this article addresses is specific. Three relevant literatures have matured separately, and none of them answers the efficacy question on its own. The higher education marketing literature has repeatedly been diagnosed as theoretically thin and weakly connected to the strategies institutions actually deploy (Hemsley-Brown & Oplatka, 2006), and its branding stream concentrates on identity and internal engagement rather than on promotional formats (Chapleo, 2010; Wæraas & Solbakk, 2009). The sustainability-in-higher-education literature examines implementation, reporting, and impact measurement, but treats communication mainly as disclosure rather than persuasion (Ceulemans et al., 2015; Findler et al., 2019; Fonseca et al., 2011). The green and disruptive marketing literatures contain the strongest causal evidence on promotional efficacy and its risks, but that evidence comes almost entirely from consumer-goods and retail contexts (Dinh & Mai, 2016; Hutter & Hoffmann, 2014; Parguel et al., 2011). No published synthesis, to the author's knowledge, integrates these strands to specify which disruptive promotional strategies have documented efficacy for sustainability messages in higher education, under what credibility conditions, and with what exposure to greenwashing perceptions. The gap matters for university marketing leaders who are already spending on such campaigns without an evidence map, for institutional leaders accountable for the integrity of sustainability claims, and for researchers who need a structured agenda rather than scattered adjacent findings. Accordingly, this article pursues three research questions. RQ1: Which disruptive promotional strategies show documented efficacy relevant to marketing sustainable practices in higher education, and at what outcome levels? RQ2: Under what credibility conditions does such promotion support, rather than undermine, prospective-student outcomes? RQ3: What risks, particularly greenwashing perceptions, do these strategies carry for universities, and how can they be mitigated? The article contributes an integrative synthesis across the three literatures, a typology of disruptive promotional strategies with their documented outcomes and evidentiary contexts, and a conceptual framework that positions credibility conditions as the mediating layer between promotional format and student-facing outcomes. The framework is a conceptual contribution derived from the synthesis; it is proposed for empirical testing, not presented as a tested model. The article proceeds as follows. The next section reviews the three literatures thematically and builds toward the gap. The Method section sets out the integrative review protocol. The Findings section answers the research questions in four themed subsections, including the typology (Table 1) and the framework (Figure 1). The Discussion develops theoretical and practical implications, acknowledges limitations, states how far the gap has been closed, and outlines future research. A brief Conclusion follows. 2. Literature Review 2.1.Marketing and Branding in the Higher Education Context Systematic reviewing of higher education marketing began from a sobering diagnosis: Hemsley-Brown and Oplatka (2006) found the literature incoherent, short of theoretical models fitted to the specific character of higher education services, and largely silent on the strategies institutions actually implement. Subsequent branding research substantiated the difficulty of importing corporate techniques. Chapleo (2010) reported that even United Kingdom universities regarded as successful brand builders struggled with internal brand engagement, and that empirical knowledge about branding objectives and measurement was scarce. Wæraas and Solbakk (2009) documented, in a Norwegian case, that the consistency and precision demanded by branding provoked internal resistance, and argued that universities may be too internally plural for a single brand essence. Website-based brand communication has likewise been questioned for its effectiveness (Chapleo et al., 2011). Taken together, this stream establishes two premises for the present review: universities operate under branding constraints that commercial advertisers do not face, and claims projected outward must survive contestation from internal constituencies who regard themselves as custodians of institutional identity. On the demand side, prospective students are active information seekers rather than passive advertising audiences. Survey evidence from 1,641 Portuguese applicants placed the university website among the three most used information sources and identified geographic proximity as a significant choice factor (Simões & Soares, 2010). Analysis of 865 questions posted by prospective students on a question-and-answer platform found that information seeking clustered on reputation, career prospects, learning and teaching, administration, and student life (Le et al., 2019). Notably, sustainability did not surface as a named dimension in that inventory, which cautions against assuming that green credentials currently dominate choice deliberations. Segmentation research on Dutch secondary students similarly concluded that, although social media penetration was very high, its influence on institutional choice was modest relative to conventional channels (Constantinides & Zinck Stagno, 2011). 2.2.Sustainability as Institutional Practice and as Communicative Resource The sustainability-in-higher-education literature charts a sector that has embraced commitments faster than it has built verification capacity. A worldwide survey-based review documented the breadth of universities' commitment to and implementation of sustainable development (Lozano et al., 2015), while exploratory work in Portugal catalogued persistent barriers and challenges to becoming a sustainable institution (Aleixo et al., 2018). Reviewing studies published between 2005 and 2017, Findler et al. (2019) identified six areas in which universities affect sustainable development but reported a preponderance of single-case studies and a gap in holistic impact assessment. The accountability infrastructure is similarly immature. Fonseca et al. (2011) examined the reports of the 25 largest Canadian universities and found sustainability reporting to be an uncommon and heterogeneous practice, limited in scope, oriented to eco-efficiency, and of limited value for decision making; Ceulemans et al. (2015) reached a compatible conclusion in a comprehensive literature review and set a research agenda for the field. Rankings intended to make campus sustainability comparable, such as GreenMetric, have themselves been subjected to critical appraisal (Lauder et al., 2015). One strand does suggest communicative upside: stakeholders of a large Romanian university, including students and staff, expressed highly favourable perceptions of the institution's sustainability orientation (Dabija et al., 2017). The pattern relevant here is asymmetry: sustainability supplies attractive message content, while the substantiation systems that would anchor aggressive promotion of that content remain uneven. 2.3.Disruptive Promotion: Mechanisms and Documented Effects The term disruptive promotional strategies is used in this article, as a conceptual category, for formats whose persuasive mechanism depends on violating audience expectations about where, when, or how promotional messages appear: guerrilla and ambient interventions, viral campaign content, experiential events, and cause-led advocacy executions. The mechanism evidence is clearest for surprise and creativity. Hutter and Hoffmann (2014) modelled ambient media effectiveness in retailing around the elicitation of surprise. Dinh and Mai (2016), studying Generation Y responses to 20 guerrilla advertisements, found that creativity, understood as the combination of novelty and relevance, had the strongest direct and indirect effects on word-of-mouth intention, that surprise contributed significantly, and that message credibility mediated these effects. For virality, Akpinar and Berger (2017) demonstrated with advertising data and experiments that emotional appeals are shared more, while informative appeals do more for brand evaluations, and that emotional content with integral brand placement combines both benefits. For experiential formats, Zarantonello and Schmitt (2013) traced the effect of event marketing on brand equity through brand experience and brand attitude. Higher-education-specific evidence concentrates on social media. Rutter et al. (2016), analysing 56 UK universities, found that social media activity related positively to brand and recruitment performance and that the association was strongest when institutions used the platforms interactively rather than as broadcast channels; their results also suggested that engagement could partially offset weaker prior reputation. Content analysis of the Facebook presences of 66 leading US institutions showed that engagement varied significantly with post type and institution type, identifying content characteristics associated with higher engagement (Peruta & Shields, 2017). Bélanger et al. (2014) documented how Canadian universities deploy social media for branding. Against these positive indications stand the avoidance findings of Kelly et al. (2010) and the modest choice influence reported by Constantinides and Zinck Stagno (2011), which jointly imply that engagement metrics should not be read as enrolment effects. 2.4.Green Marketing, Greenwashing, and the Credibility Problem Green marketing scholarship supplies the strongest warnings. Peattie and Crane (2005) argued that much of what passed for green marketing consisted of five false variants, including green spinning and green selling, lacking genuine environmental or marketing substance. The greenwashing literature has since specified drivers, forms, and effects. Delmas and Burbano (2011) located the drivers of greenwashing at institutional, market, organisational, and individual levels; Lyon and Montgomery (2015) synthesised the field and characterised greenwash as an umbrella term covering many varieties of misleading environmental communication; de Freitas Netto et al. (2020) organised those varieties into firm-level versus product-level and claim versus executional forms. Effects research is experimentally grounded. Nyilasy et al. (2014), in an experiment with 302 participants, found that when a firm's environmental performance was low, green advertising produced worse brand attitudes than general advertising or none, and that even high performers could suffer relative to silence, a pattern the authors explained through attribution of self-serving motives. Schmuck et al. (2018), in experiments in the United States (N = 486) and Germany (N = 300), found that false claims raised perceived greenwashing and damaged attitudes, that vague claims frequently escaped detection, and that nature imagery generated positive affect capable of outweighing consumers' recognition of greenwashing. Chen and Chang (2013) showed that perceived greenwash erodes green trust directly and through consumer confusion and perceived risk, and Parguel et al. (2011) demonstrated that independent sustainability ratings discipline the inferences audiences draw from responsibility communication. Morhart et al. (2015) provide the positive counterpart: perceived brand authenticity, comprising credibility, integrity, symbolism, and continuity, strengthens attachment and word of mouth. Two general theories knit these strands together. Signalling theory treats promotion as information transmitted under asymmetry, with receivers assessing signal honesty and cost (Connelly et al., 2011). Legitimacy theory treats organisational claims as bids for a perception of appropriateness that audiences confer and can withdraw (Suchman, 1995). Both predict that promotional formats which increase the visibility of a claim also increase the price of its falsification. What none of the reviewed literatures provides is an integration: the higher education marketing stream evaluates channels without sustainability content, the sustainability stream evaluates substance without promotion, and the green marketing stream evaluates promotion without the higher education context. The present review was designed to close that integration gap to the extent the published evidence allows. 3. Method 3.1.Design This study is an integrative literature review with a conceptual contribution, following the methodological guidance of Torraco (2005) and Whittemore and Knafl (2005), and the review typology of Snyder (2019). The integrative form was chosen because the question spans literatures that differ in method and maturity, and because integrative reviews are suited to synthesising diverse designs and to generating frameworks rather than pooled effect estimates (Torraco, 2005; Whittemore & Knafl, 2005). The paper evaluates the efficacy of disruptive promotional strategies by synthesising existing published evidence; no new empirical data were collected, and no surveys, interviews, or experiments were conducted by the author. 3.2.Search Strategy Literature was identified through structured searches of Scopus, Web of Science, and Google Scholar, complemented by backward and forward citation chasing from key reviews and by bibliographic verification against the Crossref database. Example search strings, adapted to each database's syntax, included: ("higher education" OR universit*) AND (marketing OR branding OR recruitment) AND (sustainab* OR green OR environment*); (guerrilla OR ambient OR viral OR experiential) AND (marketing OR advertis*) AND (effect* OR efficacy OR response); greenwash* AND (consumer OR student OR trust OR attitude); and ("higher education" OR campus) AND (sustainability reporting OR sustainability ranking). The core search window was 2005 to 2025, a period that covers the maturation of higher education marketing scholarship after the review by Hemsley-Brown and Oplatka (2006) and of greenwashing research after Peattie and Crane (2005). Seminal earlier theory was retained where the synthesis required it (Suchman, 1995). One institutional policy document (UNESCO, 2020) was included for policy context. 3.3.Inclusion and Exclusion Criteria Sources were included if they (a) were peer-reviewed journal articles, or named institutional policy documents used solely for context; (b) were published in English; (c) addressed at least one of the three review domains: higher education marketing, branding, or student choice; sustainability implementation, reporting, or perception in higher education; or the mechanisms, efficacy, or risks of disruptive or environmental promotional formats; and (d) reported empirical findings, systematic syntheses, or explicit theory development. Sources were excluded if they were practitioner commentary without method; if they concerned sustainability teaching outcomes with no communicative or marketing relevance; if their bibliographic metadata could not be verified against the publisher record; or if they duplicated findings already captured by a stronger design in the same stream. Screening proceeded in two stages, first on titles and abstracts against the domain criteria, then on full records against the evidentiary criteria. Thirty-seven sources satisfied all criteria and form the evidence base. 3.4.Appraisal and Synthesis Because the included designs range from experiments to content analyses and conceptual reviews, efficacy evidence was appraised with a design-sensitive logic rather than a single quality score. Causal claims were accepted only from experimental studies (e.g., Nyilasy et al., 2014; Parguel et al., 2011; Schmuck et al., 2018); associational claims from modelled survey or archival studies were reported as associations (e.g., Rutter et al., 2016); descriptive claims from content analyses were confined to description (e.g., Le et al., 2019; Peruta & Shields, 2017). Each finding was additionally tagged for contextual origin, distinguishing higher education samples from consumer-goods and retail samples, so that transferability could be argued rather than assumed. Synthesis followed the iterative juxtaposition procedure recommended for integrative reviews (Torraco, 2005): strategy types, outcome levels, enabling conditions, and risk mechanisms were coded from the included studies, compared across streams, and assembled into the typology reported in Table 1 and the framework in Figure 1. The framework is derived from, and traceable to, the coded evidence, but it remains a conceptual synthesis awaiting empirical test. 3.5.Rigor and Trustworthiness Several safeguards address the known weaknesses of integrative reviewing. All bibliographic records, including authors, year, outlet, pagination, and DOI, were verified against Crossref or publisher records before inclusion, and every substantive claim attributed to a source was checked against that source's published abstract or record. Attribution strength was calibrated to design, as described above, and single studies are not presented as consensus. The review does not claim the exhaustiveness of a systematic review, does not report screening flow counts, and applies no meta-analytic pooling, because the heterogeneity of designs and outcomes would make pooled estimates misleading. Scope boundaries are explicit: the review concerns promotion to prospective students in higher education; it does not evaluate sustainability curricula, campus operations, or staff-facing communication except as substantiation for external claims. Residual risks of selection and interpretation bias inherent in single-author integrative reviews are acknowledged in the Limitations subsection. 4. Findings 4.1.A Typology of Disruptive Promotional Strategies and Their Documented Efficacy The synthesis identifies four strategy families relevant to marketing sustainable practices in higher education, summarised with their documented outcomes in Table 1. The first family, guerrilla and ambient interventions, stakes persuasion on surprise encountered in unexpected physical settings. Its efficacy evidence is consistent at the level of psychological response and word of mouth: surprise is the operative emotional mechanism in ambient promotion (Hutter & Hoffmann, 2014), and creativity combining novelty with relevance exerts the strongest influence on word-of-mouth intention, with message credibility mediating the effect (Dinh & Mai, 2016). No included study tested guerrilla formats with university sponsors or sustainability content; the efficacy attributed to this family in higher education is therefore an argued transfer, not a documented result. The second family, viral and social media campaigning, carries the strongest higher-education evidence. Emotional appeals travel further than informative ones, while informative appeals do more for brand evaluations, and the two can be combined through integral brand placement (Akpinar & Berger, 2017). Within the sector, social media activity across 56 UK universities was positively associated with brand and recruitment performance, with the strongest association when platforms were used interactively (Rutter et al., 2016), and engagement with institutional Facebook content varied systematically with post and institution type across 66 leading US institutions (Peruta & Shields, 2017). Two qualifications restrain enthusiasm. Adolescent audiences actively avoid social network advertising they find irrelevant or untrustworthy (Kelly et al., 2010), and survey evidence indicates that social media influence on actual institutional choice is modest relative to conventional channels (Constantinides & Zinck Stagno, 2011). The documented efficacy of this family is therefore real but concentrated upstream of enrolment. The third family, experiential and event-based formats, is supported by evidence that event marketing builds brand equity through brand experience and brand attitude (Zarantonello & Schmitt, 2013). Campus open days and sustainability-themed events are natural higher-education vehicles, and experiential formats have the advantage of displaying practices rather than asserting them; direct sector-specific efficacy studies are nonetheless absent from the included evidence. The fourth family, cause-led advocacy communication, places the institution's sustainability commitments at the centre of the message. Here the green advertising evidence applies most directly and is conditional in sign: green appeals improved nothing when claimed performance was doubted, and depressed brand attitudes when performance was low (Nyilasy et al., 2014), while false claims raised perceived greenwashing and damaged evaluations (Schmuck et al., 2018). In answer to RQ1: documented efficacy exists chiefly for engagement-level and attitudinal outcomes, is strongest and most sector-specific for interactive social media use, and is nowhere yet documented for enrolment as an outcome of disruptive sustainability promotion. Table 1 Typology of Disruptive Promotional Strategies for Sustainability Messages in Higher Education: Mechanisms, Documented Outcomes, and Risks Strategy family Core mechanism Documented outcomes (context) Principal risk Key sources Guerrilla and ambient interventions Surprise from schema violation; creativity as novelty plus relevance Word-of-mouth intention, mediated by message credibility (consumer samples) Credibility loss when message is unclear or stunt overshadows claim Hutter & Hoffmann (2014); Dinh & Mai (2016) Viral and social media campaigns Emotional arousal drives sharing; interactivity builds relationships Sharing and brand evaluations (advertising data); association with university brand and recruitment performance; engagement varying by content type (HE samples) Advertising avoidance under irrelevance or scepticism; weak link to final choice Akpinar & Berger (2017); Rutter et al. (2016); Peruta & Shields (2017); Kelly et al. (2010); Constantinides & Zinck Stagno (2011) Experiential and event-based formats Direct brand experience shaping attitude Brand equity via brand experience and brand attitude (consumer samples) Inflated expectations exceeding campus reality Zarantonello & Schmitt (2013); Morhart et al. (2015) Cause-led advocacy communication Value alignment; moral positioning of institution Conditional: attitudes improve only when claims match performance; false claims damage attitudes; third-party ratings discipline inferences (experimental consumer samples) Greenwashing perception, trust erosion, motive attribution Nyilasy et al. (2014); Schmuck et al. (2018); Parguel et al. (2011); Chen & Chang (2013) Note. Strategy families and cell contents are derived from the studies cited in the final column; outcome descriptions preserve the original study contexts, and no cell reports evidence beyond what those sources document. HE = higher education. Consumer-context findings are transferable to higher education only as argued in the text. 4.2.Credibility Conditions: When Disruption Helps RQ2 asks under what conditions disruptive promotion of sustainable practices supports prospective-student outcomes. The synthesis supports a signalling reading (Connelly et al., 2011): prospective students operate under information asymmetry and actively seek reputational and experiential information (Le et al., 2019; Simões & Soares, 2010), so promotional formats function as signals whose value depends on perceived honesty rather than on volume. Four credibility conditions recur across the evidence. First, claim-performance alignment: the experimental green-marketing results are unambiguous that promotion detached from verifiable performance is worse than restraint (Nyilasy et al., 2014), and that outright false claims are reliably punished when detected (Schmuck et al., 2018). Second, substantiation and external verification: independent sustainability ratings changed the motives audiences attributed to responsibility communication (Parguel et al., 2011), which in the university context assigns instrumental value to credible reporting and audited indicators over self-declared virtue. Third, authentic voice: perceived brand authenticity, resting on credibility, integrity, symbolism, and continuity, strengthens attachment and word of mouth (Morhart et al., 2015); in universities, authenticity has an internal precondition, since branding that faculty and students do not recognise as true provokes resistance rather than advocacy (Chapleo, 2010; Wæraas & Solbakk, 2009). Fourth, message-channel fit: interactivity, not mere presence, carried the association with recruitment performance in the sector evidence (Rutter et al., 2016), and content characteristics measurably conditioned engagement (Peruta & Shields, 2017). Where these four conditions hold, the attention premium of disruptive formats can plausibly be converted into attitudinal and word-of-mouth gains; where they fail, the same premium accelerates the damage. 4.3.The Substantiation Deficit in Higher Education The credibility conditions expose a sector-specific vulnerability. The sustainability-in-higher-education literature shows commitments outpacing verification: implementation is uneven across institutions (Aleixo et al., 2018; Lozano et al., 2015), impact measurement is fragmented and rarely holistic (Findler et al., 2019), and reporting, where it exists, has been found uncommon, heterogeneous, narrow in scope, and of limited decision value (Fonseca et al., 2011), a diagnosis echoed at review level (Ceulemans et al., 2015). Sustainability rankings could in principle supply the third-party discipline that Parguel et al. (2011) found effective, but the most visible global campus ranking has itself drawn methodological criticism (Lauder et al., 2015), which weakens its verifying force. This is an interpretive judgement rather than a reported finding, but it follows directly from the juxtaposition: many universities currently promote from a substantiation base that could not withstand the scrutiny that disruptive visibility invites. The favourable stakeholder perceptions documented where sustainability orientation is well established (Dabija et al., 2017) indicate the upside available to institutions that close this deficit first. In legitimacy terms (Suchman, 1995), disruptive sustainability promotion is a high-variance legitimacy bid: it can accelerate legitimacy gains for institutions with substance, and accelerate legitimacy loss for institutions without it. 4.4.Risks, Reversal, and the Conceptual Framework RQ3 concerns risks. The greenwashing literature maps them precisely onto the university case. The drivers identified at institutional, market, organisational, and individual levels (Delmas & Burbano, 2011) all have higher-education analogues: ranking and league-table pressure, intensified recruitment competition, decentralised communication offices, and optimistic campaign cultures. The forms catalogued by de Freitas Netto et al. (2020) transfer as well; a university may greenwash at claim level, by overstating carbon or curricular commitments, or at executional level, through imagery that implies environmental virtue without asserting it. The executional route deserves emphasis because Schmuck et al. (2018) found that nature imagery generated positive affect that could outweigh audiences' recognition of greenwashing; a visually spectacular ambient stunt could thus succeed affectively while failing ethically, exposing the institution to delayed reputational cost when the gap surfaces. Perceived greenwash, once triggered, erodes trust both directly and through confusion and perceived risk (Chen & Chang, 2013), and the broader trajectory of greenwashing scholarship documents rising public vigilance (Lyon & Montgomery, 2015). The five false green marketings described by Peattie and Crane (2005) read, two decades later, as a checklist of failure modes for university campaigns assembled around content rather than practice. Figure 1 integrates the synthesis into a conceptual framework. Strategy families, at the left, do not act on outcomes directly; their effects pass through the credibility conditions identified above. When conditions are met, pathways lead to attention and engagement, brand attitude and trust, enrolment-related preference, and legitimacy. When conditions fail, the framework routes effects through greenwashing perception risk, which attenuates or reverses outcome gains, consistent with the experimental evidence of backfire (Nyilasy et al., 2014; Schmuck et al., 2018). A feedback path returns reputational outcomes to subsequent campaign choices, reflecting the iterative character of institutional legitimacy (Suchman, 1995). The framework's propositions are testable: each arrow specifies a relationship that higher-education-specific research can operationalise. Figure 1. Conceptual framework of credibility-conditioned efficacy of disruptive promotional strategies for sustainability messages in higher education. Solid arrows denote proposed enabling pathways; dashed arrows denote risk and feedback pathways. Source: author's synthesis of the reviewed literature. 5. Discussion 5.1.Theoretical Implications The synthesis carries three theoretical implications. First, it relocates the efficacy question. The literatures reviewed do not support asking whether disruptive promotional strategies work in general; the defensible question is at which outcome level, under which credibility conditions, and at what risk. Treating efficacy as conditional aligns higher education marketing with the attribution-theoretic turn in green advertising research (Nyilasy et al., 2014; Parguel et al., 2011) and answers, for one domain, the call to theorise mechanisms of misleading environmental communication (Lyon & Montgomery, 2015). Second, the review extends greenwashing theory into a public-sector service context in which the claimant is simultaneously an educator of the audiences that will judge it, a reflexivity absent from consumer-goods settings and likely to intensify scrutiny; this extension is conceptual and awaits empirical confirmation. Third, by joining signalling (Connelly et al., 2011), legitimacy (Suchman, 1995), and authenticity (Morhart et al., 2015) perspectives around a single promotional problem, the framework specifies how internal brand contestation, documented as a distinctive university condition (Wæraas & Solbakk, 2009), becomes an externally visible credibility variable rather than a private management difficulty. 5.2.Practical Implications For university marketing leaders, the evidence ordering is substance first, verification second, spectacle third. Institutions should audit the substantiation base, including reporting quality and externally checkable indicators, before commissioning high-visibility sustainability campaigns, because the experimental record shows promotion without performance to be value-destroying rather than merely ineffective (Nyilasy et al., 2014). Claims should be specific and checkable in preference to vague, since vague claims that initially escape detection (Schmuck et al., 2018) leave the institution holding latent risk. Third-party anchors, such as audited reports or credible rankings, should accompany campaign claims where available (Parguel et al., 2011), with awareness of the methodological contestation surrounding some rankings (Lauder et al., 2015). Within the disruptive repertoire, the strongest sector evidence favours interactive social media engagement over broadcast spectacle (Rutter et al., 2016), and content planning should reflect measured engagement differences by content type (Peruta & Shields, 2017). Campaigns fronted by verifiable student experience, for example experiential events that display actual campus practices, fit both the experiential evidence (Zarantonello & Schmitt, 2013) and the authenticity requirement (Morhart et al., 2015). Finally, expectations should be calibrated: no included study links disruptive sustainability promotion to enrolment, and information-seeking evidence suggests sustainability is not yet a dominant choice dimension (Le et al., 2019), so such campaigns are better justified as brand and legitimacy investments than as direct recruitment instruments. 5.3.Limitations The limitations follow from the method. An integrative review is selective by design; despite structured searches, relevant studies may have been missed, and the restriction to English-language, peer-reviewed sources introduces language and publication bias. The evidence base for two of the four strategy families rests on consumer and retail samples, so the transfers argued here, however disciplined, remain inferences rather than findings. Heterogeneity of designs precluded meta-analytic pooling, and no screening flow counts are reported, so the review should not be read as exhaustive in the systematic-review sense. Coding and framework derivation were performed by a single author, which admits interpretation bias that the traceability of every claim to logged, verified sources mitigates but cannot eliminate. Finally, the framework is a conceptual synthesis; none of its pathways has been tested as an integrated model. 5.4.How Far the Gap Was Closed, and Future Research The integration gap identified in the Introduction has been closed at the level of synthesis: the review now connects the three literatures, specifies where documented efficacy exists and where it is absent, and converts scattered warnings into a structured risk mechanism. The gap remains open at the level of causal evidence, because higher-education-specific tests of disruptive sustainability promotion do not yet exist in the included record. Future research should proceed on four fronts: experimental studies manipulating claim specificity, substantiation, and format within realistic university recruitment stimuli; longitudinal designs linking campaign exposure to application and enrolment behaviour rather than engagement proxies; measurement work adapting greenwashing-perception and brand-authenticity scales (Chen & Chang, 2013; Morhart et al., 2015) to institutional claims; and comparative studies across national systems, since choice factors and media behaviour differ across the contexts represented in the evidence base (Constantinides & Zinck Stagno, 2011; Simões & Soares, 2010). 6. Conclusion This article evaluated the efficacy of disruptive promotional strategies for marketing sustainable practices in higher education by integrating evidence from higher education marketing, sustainability in higher education, and green and disruptive marketing research. The evidence supports a restrained verdict. Disruptive formats have documented power to generate attention, engagement, and word of mouth, and interactive social media use shows sector-specific associations with brand and recruitment performance; nothing in the record yet ties these formats to enrolment outcomes for sustainability messages. Their efficacy is conditional on credibility: alignment between claim and performance, external substantiation, authentic institutional voice, and fit between message and channel. Because the sector's verification infrastructure remains uneven, the same properties that make these strategies attractive make them hazardous for institutions whose practice lags their positioning. The typology and conceptual framework contributed here organise what is known, mark the boundary of what is documented, and specify the tests that would move the field from argued transferability to sector-specific evidence. Declarations Funding: This research received no external funding. Conflicts of Interest: The author declares no conflict of interest. Ethics: This study is a review of published literature and did not involve human participants, animals, or personal data; ethical approval was therefore not required. Data Availability: No new data were created or analysed in this study. All sources synthesised are publicly available through the cited references. References Akpinar, E., & Berger, J. (2017). Valuable virality. Journal of Marketing Research, 54(2), 318–330. https://doi.org/10.1509/jmr.13.0350 Aleixo, A. M., Leal, S., & Azeiteiro, U. M. (2018). Conceptualization of sustainable higher education institutions, roles, barriers, and challenges for sustainability: An exploratory study in Portugal. Journal of Cleaner Production, 172, 1664–1673. https://doi.org/10.1016/j.jclepro.2016.11.010 Bélanger, C. H., Bali, S., & Longden, B. (2014). How Canadian universities use social media to brand themselves. Tertiary Education and Management, 20(1), 14–29. https://doi.org/10.1080/13583883.2013.852237 Ceulemans, K., Molderez, I., & Van Liedekerke, L. (2015). Sustainability reporting in higher education: A comprehensive review of the recent literature and paths for further research. Journal of Cleaner Production, 106, 127–143. https://doi.org/10.1016/j.jclepro.2014.09.052 Chapleo, C. (2010). What defines "successful" university brands? International Journal of Public Sector Management, 23(2), 169–183. https://doi.org/10.1108/09513551011022519 Chapleo, C., Carrillo Durán, M. V., & Castillo Díaz, A. (2011). Do UK universities communicate their brands effectively through their websites? Journal of Marketing for Higher Education, 21(1), 25–46. https://doi.org/10.1080/08841241.2011.569589 Chen, Y.-S., & Chang, C.-H. (2013). Greenwash and green trust: The mediation effects of green consumer confusion and green perceived risk. Journal of Business Ethics, 114(3), 489–500. https://doi.org/10.1007/s10551-012-1360-0 Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling theory: A review and assessment. Journal of Management, 37(1), 39–67. https://doi.org/10.1177/0149206310388419 Constantinides, E., & Zinck Stagno, M. C. (2011). Potential of the social media as instruments of higher education marketing: A segmentation study. Journal of Marketing for Higher Education, 21(1), 7–24. https://doi.org/10.1080/08841241.2011.573593 Dabija, D.-C., Postelnicu, C., Dinu, V., & Mihăilă, A. (2017). Stakeholders' perception of sustainability orientation within a major Romanian university. International Journal of Sustainability in Higher Education, 18(4), 533–553. https://doi.org/10.1108/IJSHE-10-2015-0169 de Freitas Netto, S. V., Sobral, M. F. F., Ribeiro, A. R. B., & Soares, G. R. L. (2020). Concepts and forms of greenwashing: A systematic review. Environmental Sciences Europe, 32(1), Article 19. https://doi.org/10.1186/s12302-020-0300-3 Delmas, M. A., & Burbano, V. C. (2011). The drivers of greenwashing. California Management Review, 54(1), 64–87. https://doi.org/10.1525/cmr.2011.54.1.64 Dinh, T. D., & Mai, K. N. (2016). Guerrilla marketing's effects on Gen Y's word-of-mouth intention: A mediation of credibility. Asia Pacific Journal of Marketing and Logistics, 28(1), 4–22. https://doi.org/10.1108/APJML-06-2015-0102 Findler, F., Schönherr, N., Lozano, R., Reider, D., & Martinuzzi, A. (2019). The impacts of higher education institutions on sustainable development: A review and conceptualization. International Journal of Sustainability in Higher Education, 20(1), 23–38. https://doi.org/10.1108/IJSHE-07-2017-0114 Fonseca, A., Macdonald, A., Dandy, E., & Valenti, P. (2011). The state of sustainability reporting at Canadian universities. International Journal of Sustainability in Higher Education, 12(1), 22–40. https://doi.org/10.1108/14676371111098285 Hemsley-Brown, J., & Oplatka, I. (2006). Universities in a competitive global marketplace: A systematic review of the literature on higher education marketing. International Journal of Public Sector Management, 19(4), 316–338. https://doi.org/10.1108/09513550610669176 Hutter, K., & Hoffmann, S. (2014). Surprise, surprise. Ambient media as promotion tool for retailers. Journal of Retailing, 90(1), 93–110. https://doi.org/10.1016/j.jretai.2013.08.001 Kelly, L., Kerr, G., & Drennan, J. (2010). Avoidance of advertising in social networking sites: The teenage perspective. Journal of Interactive Advertising, 10(2), 16–27. https://doi.org/10.1080/15252019.2010.10722167 Lauder, A., Sari, R. F., Suwartha, N., & Tjahjono, G. (2015). Critical review of a global campus sustainability ranking: GreenMetric. Journal of Cleaner Production, 108, 852–863. https://doi.org/10.1016/j.jclepro.2015.02.080 Le, T. D., Dobele, A. R., & Robinson, L. J. (2019). Information sought by prospective students from social media electronic word-of-mouth during the university choice process. Journal of Higher Education Policy and Management, 41(1), 18–34. https://doi.org/10.1080/1360080X.2018.1538595 Lozano, R., Ceulemans, K., Alonso-Almeida, M., Huisingh, D., Lozano, F. J., Waas, T., Lambrechts, W., Lukman, R., & Hugé, J. (2015). A review of commitment and implementation of sustainable development in higher education: Results from a worldwide survey. Journal of Cleaner Production, 108, 1–18. https://doi.org/10.1016/j.jclepro.2014.09.048 Lyon, T. P., & Montgomery, A. W. (2015). The means and end of greenwash. Organization & Environment, 28(2), 223–249. https://doi.org/10.1177/1086026615575332 Morhart, F., Malär, L., Guèvremont, A., Girardin, F., & Grohmann, B. (2015). Brand authenticity: An integrative framework and measurement scale. Journal of Consumer Psychology, 25(2), 200–218. https://doi.org/10.1016/j.jcps.2014.11.006 Nyilasy, G., Gangadharbatla, H., & Paladino, A. (2014). Perceived greenwashing: The interactive effects of green advertising and corporate environmental performance on consumer reactions. Journal of Business Ethics, 125(4), 693–707. https://doi.org/10.1007/s10551-013-1944-3 Parguel, B., Benoît-Moreau, F., & Larceneux, F. (2011). How sustainability ratings might deter "greenwashing": A closer look at ethical corporate communication. Journal of Business Ethics, 102(1), 15–28. https://doi.org/10.1007/s10551-011-0901-2 Peattie, K., & Crane, A. (2005). Green marketing: Legend, myth, farce or prophesy? Qualitative Market Research: An International Journal, 8(4), 357–370. https://doi.org/10.1108/13522750510619733 Peruta, A., & Shields, A. B. (2017). Social media in higher education: Understanding how colleges and universities use Facebook. Journal of Marketing for Higher Education, 27(1), 131–143. https://doi.org/10.1080/08841241.2016.1212451 Rutter, R., Roper, S., & Lettice, F. (2016). Social media interaction, the university brand and recruitment performance. Journal of Business Research, 69(8), 3096–3104. https://doi.org/10.1016/j.jbusres.2016.01.025 Schmuck, D., Matthes, J., & Naderer, B. (2018). Misleading consumers with green advertising? An affect–reason–involvement account of greenwashing effects in environmental advertising. Journal of Advertising, 47(2), 127–145. https://doi.org/10.1080/00913367.2018.1452652 Simões, C., & Soares, A. M. (2010). Applying to higher education: Information sources and choice factors. Studies in Higher Education, 35(4), 371–389. https://doi.org/10.1080/03075070903096490 Snyder, H. (2019). Literature review as a research methodology: An overview and guidelines. Journal of Business Research, 104, 333–339. https://doi.org/10.1016/j.jbusres.2019.07.039 Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20(3), 571–610. https://doi.org/10.2307/258788 Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367. https://doi.org/10.1177/1534484305278283 UNESCO. (2020). Education for sustainable development: A roadmap. UNESCO. https://unesdoc.unesco.org/ark:/48223/pf0000374802 Wæraas, A., & Solbakk, M. N. (2009). Defining the essence of a university: Lessons from higher education branding. Higher Education, 57(4), 449–462. https://doi.org/10.1007/s10734-008-9155-z Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x Zarantonello, L., & Schmitt, B. H. (2013). The impact of event marketing on brand equity: The mediating roles of brand experience and brand attitude. International Journal of Advertising, 32(2), 255–280. https://doi.org/10.2501/IJA-32-2-255-280 Hashtags: #SDG12 #ResponsibleConsumption #SDG4 #QualityEducation #SustainabilityCommunication #HigherEducationMarketing #GuerrillaMarketing #Greenwashing #UniversityBranding #StudentRecruitment #IntegrativeReview #DisruptivePromotion #U7YJournal #U7Y #AcademicResearch

Author: Maya Patel Affiliation: Swiss International University (SIU) ORCID ID: 0009-0003-9913-3240 Submitted 07 April 2026; Revised 09 June 2026; Accepted 16 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10009 Volume 3, December 2026, (SpecSDG10009) Abstract Cities are investing heavily in data platforms, sensing infrastructure, and algorithmic tools, yet the officials who must govern these systems are rarely educated for the task. This article addresses how universities should design executive academic programs that build the governance and planning capabilities smart city transformation actually requires. It proceeds as an integrative review of two literatures that have developed in isolation: research on smart city governance, which documents the competencies and organisational capacities cities lack, and research on executive education for the public sector, which documents how experienced professionals learn. From the first literature the article derives five competency domains: socio-technical governance literacy, data governance and ethics, technology appraisal for planning, collaborative innovation, and organisational transformation. From the second it derives four design principles: problem-anchored action learning, cohort composition as curriculum, co-production with employing governments, and multi-level evaluation of effects. These elements are integrated into a program design framework that maps competency domains onto curricular structures and connects program outcomes to the organisational capacities documented in the smart city literature. The framework's central claim is that executive programs succeed when the city itself becomes the learning environment, with live governance problems serving as the primary curricular material. The article states the framework's conceptual status plainly, identifies evaluation of program effects at the organisational level as the weakest link in the evidence, and sets an agenda for empirical testing in diverse institutional contexts. Keywords: executive education, smart city governance, urban planning, competency frameworks, public sector capacity, action learning, curriculum design 1. Introduction The governance of cities is being reshaped by data infrastructures, sensing networks, digital twins, and algorithmic decision support. A large research literature now agrees that the decisive constraint on this transformation is not technology but institutional and human capacity: smart city governance is a complex process of institutional change in which technology amplifies, rather than replaces, the need for skilled judgment (Meijer & Rodríguez Bolívar, 2016; Mora et al., 2023). Comparative case research finds that cities succeeding in digital transformation are those that deliberately built organisational capacity, through strategy, leadership, dedicated units, and collaboration arrangements, rather than those that procured the most technology (Gascó-Hernández et al., 2022). Interview studies with smart city managers across forty cities identify distinctive dynamic capabilities, sensing, seizing, innovation, integration, and empowerment, that such roles demand (Guenduez & Mergel, 2022). Where are these capabilities to come from? The professionals who govern and plan cities are mid-career: department heads, chief planners, digitalisation officers, and municipal executives who cannot return to full-time study and whose learning must connect directly to their organisations. The natural institutional response is executive academic education, the segment of university provision designed for experienced professionals. Yet the research on executive education for the public sector shows a field that is unevenly developed, concentrated in a small number of systems, and only loosely connected to the substantive challenges of digital governance (Rasmussen & Callan, 2016; Sørensen, 2023). Meanwhile, the smart city literature has produced competency checklists, capability frameworks, and workforce analyses (Kwon et al., 2025; Bastidas et al., 2024; David & McNutt, 2019) that stop short of specifying how an actual program of study should be designed, staffed, and evaluated. The scale of the need is not marginal. Nearly every city pursuing a digital agenda confronts the same staffing reality: the officials responsible for data platforms, procurement of algorithmic tools, and digitally mediated participation were educated before these instruments existed, and the professions that feed city government, planning, administration, engineering, have been slow to adjust their curricula (David & McNutt, 2019). Recruitment alone cannot close the gap, because the roles demand combinations of institutional knowledge and technical judgment that external hires rarely carry. The remaining lever is education for serving professionals, which places the design of executive programs at the centre of smart city capability policy rather than at its periphery. The gap this article addresses is the missing connection between these two bodies of knowledge. The smart city literature tells us with increasing precision what city officials need to be able to do, but not how programs should teach it; the executive education literature tells us how experienced professionals learn, but has not been applied systematically to the smart city domain. No published framework currently integrates smart-city competency research with the pedagogy of executive education into an actionable program design. Consequently, universities designing offerings in this space work from intuition, and cities purchasing them cannot judge quality. The article asks three questions. First, which competency domains does the smart city governance and planning literature identify as necessary for senior urban professionals? Second, which design principles does the executive education literature establish for programs serving experienced public professionals? Third, how can the two be integrated into a coherent framework for developing executive academic programs in this field? The contribution is a conceptual program design framework grounded in an integrative review. Section 2 reviews both literatures. Section 3 sets out the method. Section 4 develops the framework, with a synthesising table and figure. Section 5 discusses implications and limitations, and Section 6 concludes. 2. Literature Review 2.1 Smart city governance as institutional change Reviews of smart city governance converge on a socio-technical reading. An early and influential synthesis distinguished conceptions emphasising smart technology, smart people, and smart collaboration, and argued that governing the smart city means crafting new forms of collaboration through information and communication technologies rather than deploying systems (Meijer & Rodríguez Bolívar, 2016). Subsequent systematic reviews document persistent variance in definitions, components, and outcome measures (Ruhlandt, 2018), position smart governance as technology-enabled collaboration among government, citizens, and other stakeholders (Viale Pereira et al., 2018), and find that empirical evidence for sustainability benefits remains sparse and strongly context-dependent (Tomor et al., 2019). Research on developing countries adds that technology-enabled governance presupposes concurrent reforms in regulation, human capital, and basic infrastructure (Tan & Taeihagh, 2020). A recent multidimensional synthesis identifies institutional capacity, alongside digital infrastructure, data governance, and citizen engagement, as a core enabler, and institutional inertia as a core barrier (Almulhim & Yigitcanlar, 2025). From an innovation management perspective, governance remains among the most undertheorised dimensions of smart city transitions (Mora et al., 2023). The consistent implication for education is that the object of study cannot be technology alone: officials must understand institutions, collaboration, and public value creation as the medium in which technology operates. A further implication follows from the reviews' repeated finding of definitional variance (Ruhlandt, 2018): because no settled model of smart city governance exists, programs cannot teach a canon; they must instead build the analytical capacity to evaluate competing models against local conditions, a capacity closer to judgment than to knowledge transmission. 2.2 Competencies and capacities for the smart city workforce A second stream specifies what smart city work demands of people and organisations. Case comparisons of Milan, Barcelona, and Munich show that successful cities strengthened management capacity, strategy, leadership, dedicated units, and collaboration capacity, spanning public-private partnerships and multi-level government relations (Gascó-Hernández et al., 2022). The dynamic managerial capabilities identified among smart city managers, sensing opportunities, seizing them, driving innovation, integrating across silos, and empowering staff, interact with organisational readiness factors such as resource and strategic readiness (Guenduez & Mergel, 2022). Competency research has begun to formalise these demands: a Delphi study with experts produced five main competencies and fourteen sub-competencies for smart city managers, operationalised as a thirty-six item checklist intended to feed educational recommender systems (Kwon et al., 2025). A competency framework for leading urban digital innovation identifies the tasks, roles, and competencies needed to steer digital projects toward public value and to break professional silos between city managers and built-environment professionals (Bastidas et al., 2024). Analyses of the planning and administrative professions argue that professional education has not kept pace with the governance use of information technologies and must change in both content and delivery (David & McNutt, 2019). Complementary work develops infrastructure for competence management itself, including ontology-based platforms that map skill-gap diagnoses to learning pathways for municipal staff (Iatrellis et al., 2021). Applied capacity-building studies show that competency-based training can foster collaboration across departmental silos in city government (Bhagavathula et al., 2021) and have begun to define competency profiles for emerging roles such as the smart city resilience officer (Tsoutsa et al., 2024). Two features of this stream matter for program design. First, the competencies identified are predominantly integrative rather than technical: sensing opportunities across policy fields, orchestrating actors, and reading organisational readiness are capacities exercised in context, which constrains the pedagogy that can develop them. Second, the stream is methodologically young. Much of it rests on expert elicitation, single-city cases, and framework construction rather than on longitudinal observation of officials at work, so an education program built on it must treat the competency specifications as the best available map rather than settled ground truth, and must be designed to refine them through its own evaluation data. 2.3 Data, ethics, and emerging planning tools A third stream defines the substantive knowledge base such programs must carry. Data governance research argues that value creation from urban data generates ethical and legitimacy challenges that must be translated into accountability mechanisms rather than left as abstract principles (König, 2021), and that sustainable smart city development requires trusted, participatory data governance (Paskaleva et al., 2017). Reviews of the ethical debate cluster concerns into network infrastructure and surveillance, post-political governance, social inclusion, and sustainability (Ziosi et al., 2024). On the tools side, urban digital twins have moved from prototypes to city-scale infrastructure: case studies document their use for participatory planning in Herrenberg (Dembski et al., 2020) and for scenario-based planning on open 3D data in Zurich (Schrotter & Hürzeler, 2020), while systematic reviews catalogue interoperability, data quality, and governance challenges that delay implementation (Weil et al., 2023). Maturity-model research on city digital twins concludes that their governance implications, participation, trust, inclusion of marginalised groups, demand human-centric design and adaptive regulation (Haraguchi et al., 2024). For curriculum purposes, this stream supplies both content and caution: officials need enough technical literacy to appraise such tools, and enough critical literacy to govern them. 2.4 Executive education for public professionals The final stream concerns how experienced professionals learn in academic settings. Surveys of executive Master of Public Administration programs find their strength in engaging diverse cohorts of professionals in sustained, experience-linked study, and their persistent challenges in funding, organisational relevance, and demonstrating impact (Holmes, 2012). Analyses of public policy schools find executive education unevenly developed globally, with an emerging orthodoxy built on action learning, leadership-focused curricula, and practitioner-supplemented faculty (Rasmussen & Callan, 2016). A systematic review of the effects of continuing public sector management education finds documented effects concentrated at the individual level, with organisational-level evidence scarce, a finding with direct consequences for how programs should be evaluated (Sørensen, 2023). Design scholarship proposes treating executive education as a co-produced service in which participants and their organisations are active stakeholders rather than recipients (Hiedemann et al., 2017). The pedagogical core of this literature is action learning: structured work on real problems from participants' own organisations, in facilitated peer sets, with reflection as the engine of learning. Longitudinal evidence shows action learning fosters knowledge, skills, and attitudes that transfer to professional practice years later (Perusso et al., 2021). Accounts of accredited executive programs document how action learning treats participants as experts on their own problems (Ruane, 2016), requires trust and communication among learners, academics, and external contributors (Stephens & Margey, 2015), and produces learning with emotional, social, and relational dimensions that conventional teaching does not reach (Ruane & Corlett, 2024). Program-level studies show how live projects align learning with organisational goals (Conine & Peratoner, 2019), how reflection reports, learning contracts, and action-oriented theses combine rigor with organisational action (Berggren & Söderlund, 2011), and and how the design process itself benefits from action-reflection among tutors and program managers (Kelliher & Byrne, 2018). Read together, the four streams supply complementary halves of a design problem: the first three specify the competencies and content executive programs must deliver for smart city governance and planning; the fourth specifies the pedagogy through which experienced professionals can acquire them. The integration is missing. The present article constructs it. 3. Method 3.1 Design and rationale The study is an integrative review with a design-oriented conceptual contribution. The choice follows from the research questions: identifying competency domains and design principles requires organised synthesis across heterogeneous fields, education research, public administration, urban studies, and information systems, whose methods do not support a single effect-size synthesis; and constructing a program design framework is a conceptual task that must nevertheless be anchored in verifiable evidence. A systematic review was rejected because the two literatures use incommensurable outcome constructs; a purely speculative design essay was rejected because the framework's authority depends on its evidential base. 3.2 Search and selection Literature was identified between June and August 2026 through structured queries of scholarly databases (Consensus, drawing on Semantic Scholar, Scopus, and related indexes), with verification of every candidate source against its digital object identifier or publisher record. Query families covered: smart city governance reviews and frameworks; organisational capacity and managerial capabilities in city government; competency frameworks for smart city professionals; data governance and ethics in smart cities; urban digital twins in planning; executive and continuing education for public sector professionals; and action learning in executive programs. Inclusion criteria were: peer-reviewed publication; direct relevance to at least one of the two halves of the design problem; and confirmable bibliographic identity. Exclusion criteria were: purely technical papers without governance or educational implication, promotional or unverifiable sources, and duplicative preprints of published articles, for which the journal version was retained. Thirty-three sources satisfied all criteria and constitute the evidence base. 3.3 Analytical procedure The analysis proceeded in three steps. First, sources from the smart city streams were coded for the capabilities, knowledge, and dispositions they identify as necessary for governance and planning roles; codes were consolidated into candidate competency domains, merging overlapping constructs and discarding domains supported by a single source only. Second, sources from the executive education stream were coded for design features associated with documented learning effects; features were consolidated into design principles by the same rule. Third, domains and principles were integrated: for each competency domain the analysis asked which pedagogical form the executive education evidence indicates for developing it, and the resulting mappings were assembled into the program design framework presented in Section 4, checked for internal consistency and for coverage of every retained source. 3.4 Rigor and boundaries Bias was limited by including critical scholarship on smart cities alongside advocacy, by admitting only verified sources, and by requiring multi-source support for every domain and principle. The review's boundaries are explicit: it synthesises published research rather than new fieldwork; the executive education evidence derives mainly from business and general public administration contexts rather than smart city programs specifically, since the latter barely exist in the peer-reviewed record; and the framework is a design proposition whose validation requires implementation studies. These boundaries are taken up in the Limitations subsection. 4. A Program Design Framework for Executive Education in Smart City Governance and Planning The framework integrates five competency domains derived from the smart city literature with four design principles derived from the executive education literature, and maps both onto a program architecture. Figure 1 presents the framework, and Table 1 summarises the competency domains, their evidential grounding, and their curricular expression. The subsections develop each element. 4.1 Five competency domains The coding of the smart city streams yields five consolidated domains. The first is socio-technical governance literacy: understanding smart city transformation as institutional change, including the collaboration arrangements, legitimacy claims, and public value questions that technology raises but cannot answer (Meijer & Rodríguez Bolívar, 2016; Viale Pereira et al., 2018; Mora et al., 2023). The second is data governance and ethics: the ability to design accountability mechanisms for data-based value creation, to weigh privacy, surveillance, and inclusion concerns, and to institutionalise citizen-centred data practices (König, 2021; Paskaleva et al., 2017; Ziosi et al., 2024). The third is technology appraisal for planning: sufficient literacy in sensing infrastructures, digital twins, and algorithmic tools to evaluate their claims, costs, maturity, and governance implications, exemplified by the documented capabilities and limits of city digital twins (Dembski et al., 2020; Schrotter & Hürzeler, 2020; Weil et al., 2023; Haraguchi et al., 2024). The fourth is collaborative innovation: orchestrating partnerships across departments, governments, firms, and citizens, the capacity most consistently associated with successful transformation in comparative cases (Gascó-Hernández et al., 2022; Tomor et al., 2019; Bhagavathula et al., 2021). The fifth is organisational transformation: the dynamic managerial capabilities of sensing, seizing, integrating, and empowering, coupled with the ability to assess and build organisational readiness (Guenduez & Mergel, 2022; Kwon et al., 2025; Bastidas et al., 2024). The domains are distinct but interdependent; competency research warns that treating them as isolated skill lists reproduces the silos the roles must overcome (Bastidas et al., 2024; David & McNutt, 2019). 4.2 Four design principles The executive education stream yields four principles with documented support. The first is problem-anchored action learning: the primary curricular material should be live governance problems from participants' own cities, worked in facilitated peer sets, because this is the form with the strongest evidence of durable transfer to practice (Perusso et al., 2021; Ruane, 2016; Conine & Peratoner, 2019). The second is cohort composition as curriculum: deliberately diverse cohorts, across functions, professions, and cities, are themselves a learning resource, and sustained engagement of such cohorts is the documented strength of executive public administration formats (Holmes, 2012; Stephens & Margey, 2015). The third is co-production with employing governments: programs should be designed as services co-produced with participants and their organisations, aligning projects with organisational goals rather than treating the employer as absent (Hiedemann et al., 2017; Conine & Peratoner, 2019; Kelliher & Byrne, 2018). The fourth is multi-level evaluation: because documented effects of continuing public sector education concentrate at the individual level, programs must build evaluation at individual, organisational, and professional levels into their design rather than asserting organisational impact (Sørensen, 2023; Berggren & Söderlund, 2011). 4.3 The program architecture Mapping domains onto principles produces the architecture summarised in Table 1 and Figure 1. Its organising claim is that the city is the learning environment: each competency domain is developed not as a taught subject followed by application, but through structured work on the participant's own city, with academic content sequenced around the problems that work surfaces. Socio-technical governance literacy and data ethics are developed through case seminars and critical reading anchored to participants' current initiatives; technology appraisal through structured evaluation exercises on real systems, including digital twin and data platform assessments modelled on the documented cases (Schrotter & Hürzeler, 2020; Haraguchi et al., 2024); collaborative innovation through inter-city peer sets and stakeholder engagement tasks within the action learning projects; and organisational transformation through a program-long applied project, supported by learning contracts and reflection instruments of the kind documented in rigorous executive formats (Berggren & Söderlund, 2011). Assessment combines academically examined components with organisationally validated project outcomes, giving effect to the co-production principle. Faculty composition follows the same logic: academic staff carry the research base, while practitioner contributors carry contextual credibility, a mix the executive education literature treats as standard (Rasmussen & Callan, 2016). The architecture implies a deliberate sequencing across a program cycle. An entry phase establishes shared conceptual ground in socio-technical governance and diagnoses each participant's competency profile against instruments of the kind developed in the reviewed literature (Kwon et al., 2025), so that pathways can be differentiated rather than uniform (Iatrellis et al., 2021). A development phase runs the taught domains in parallel with the action learning sets, alternating short residential blocks with workplace periods so that each block processes what the preceding workplace period surfaced, a rhythm the accounts of accredited executive programs identify as the point where reflection becomes learning (Ruane, 2016; Kelliher & Byrne, 2018). A consolidation phase centres on the applied project's organisational delivery and the reflective thesis that converts experience into transferable knowledge (Berggren & Söderlund, 2011). Program length follows from the logic rather than from convention: the cycle must span enough workplace time for a real governance problem to move, which favours designs measured in semesters rather than weeks and distinguishes this architecture from the short-course formats that dominate commercial executive provision (Rasmussen & Callan, 2016). Table 1 Competency Domains for Smart City Governance and Planning: Evidence and Curricular Expression Competency domain Core content Principal evidence Curricular expression Assessment form Socio-technical governance literacy Smart city transformation as institutional change; collaboration; public value Meijer & Rodríguez Bolívar (2016); Viale Pereira et al. (2018); Mora et al. (2023) Case seminars anchored to participants' initiatives; critical reading Examined analysis of own-city governance arrangement Data governance and ethics Accountability for data-based value creation; privacy, surveillance, inclusion König (2021); Paskaleva et al. (2017); Ziosi et al. (2024) Applied data governance audit of participant's organisation Audit report with remediation plan Technology appraisal for planning Capabilities and limits of sensing, digital twins, algorithmic tools Dembski et al. (2020); Weil et al. (2023); Haraguchi et al. (2024) Structured appraisal exercises on real deployed systems Appraisal dossier with procurement recommendation Collaborative innovation Orchestrating partnerships across departments, governments, firms, citizens Gascó-Hernández et al. (2022); Tomor et al. (2019); Bhagavathula et al. (2021) Inter-city peer sets; stakeholder engagement tasks in live projects Facilitated peer review of collaboration design Organisational transformation Dynamic capabilities; organisational readiness; leading change Guenduez & Mergel (2022); Kwon et al. (2025); Bastidas et al. (2024) Program-long action learning project with learning contract Organisationally validated project outcome; reflective thesis Note. Domains were consolidated from the reviewed smart city literature; curricular expressions apply the design principles derived from the executive education literature (Section 4.2). Assessment forms combine academic examination with organisational validation, following Berggren and Söderlund (2011) and Hiedemann et al. (2017). 4.4 Evaluation and institutional embedding The framework treats evaluation as a design component rather than an afterthought. Following the evidence that individual-level effects are the best documented (Sørensen, 2023), the architecture specifies three evaluation levels: individual learning, assessed through examined components; organisational effect, assessed through the validated outcomes of action learning projects and follow-up at defined intervals; and professional-field effect, observed through alumni networks and inter-city diffusion of practices, the level on which evidence is thinnest and structured data collection most valuable. Institutionally, the framework implies that programs cannot be run as detached teaching products: they require standing partnerships with city governments for project access and validation, a faculty mix spanning research and practice, and governance arrangements that let curricula track a fast-moving field, updating technology appraisal content as tools such as digital twins mature (Weil et al., 2023). Competence management infrastructure of the kind developed for municipal workforces can support diagnostic entry assessment and personalised pathways within the program (Iatrellis et al., 2021; Kwon et al., 2025). The framework also anticipates its characteristic failure modes, each corresponding to a suppressed component. A program that retains the taught domains but drops the action learning projects reverts to the instruction-only format whose transfer to practice the evidence base gives least reason to expect (Perusso et al., 2021). A program that runs projects without organisational co-production produces academically interesting work the employing government never validates or adopts, the relevance failure long attributed to detached provision (Hiedemann et al., 2017; Rasmussen & Callan, 2016). A program that treats the technology appraisal domain as vendor-led training rather than critical appraisal reproduces inside the classroom the promotional asymmetry that already distorts municipal procurement, precisely the condition the governance literature warns against (Meijer & Rodríguez Bolívar, 2016; Haraguchi et al., 2024). Naming these failure modes converts the framework from a description of good practice into an instrument for diagnosing weak practice, which is how design frameworks earn their keep in institutional settings. Figure 1. Program design framework for executive academic programs in smart city governance and planning. Five competency domains (left) derived from the smart city literature are developed through four design principles (centre) derived from the executive education literature, yielding a program architecture (right) in which the participant's own city is the primary learning environment; evaluation feeds back at individual, organisational, and professional levels. Source: author's elaboration based on the reviewed literature. 5. Discussion 5.1 Theoretical implications The framework makes two theoretical moves. First, it recasts the smart city capability problem as an educational design problem. The competency literature has been accumulating checklists and frameworks (Kwon et al., 2025; Bastidas et al., 2024) whose implicit theory of change, that specifying competencies leads to their development, leaves the development mechanism unexamined. Connecting these frameworks to the action learning evidence supplies the missing mechanism: competencies of the kind the smart city literature describes, judgment-laden, collaborative, and organisationally embedded, are precisely those the executive education literature shows are developed through structured work on real problems rather than through instruction alone (Perusso et al., 2021; Ruane & Corlett, 2024). Second, the framework gives the executive education literature a substantive domain in which its service-dominant and co-production arguments (Hiedemann et al., 2017) acquire specific institutional content: in smart city programs, co-production is not merely good pedagogy but the only access route to the live systems, data arrangements, and stakeholder networks the curriculum requires. The framework also clarifies a boundary condition. Because empirical evidence on smart governance outcomes is context-dependent (Tomor et al., 2019) and developing-country conditions differ structurally (Tan & Taeihagh, 2020), the framework specifies design logic rather than fixed content: the domains are stable, but their content must be localised, which is itself an argument for the problem-anchored pedagogy the framework adopts. A third implication concerns the relationship between education and evidence generation. The organisational-level evidence gap that Sørensen (2023) documents is usually read as a shortcoming of evaluation research; the framework reads it instead as a consequence of program designs in which organisational effect was never a structural component and therefore never measurable. Programs built to the present architecture generate organisational-level data as a by-product of their operation, through validated project outcomes and follow-up, which means each implementation is simultaneously an intervention and a study. Executive education in this reading is not only a consumer of the smart city evidence base but one of the few available instruments for producing it, since the action learning projects constitute structured, documented interventions in real city governance of a kind researchers rarely gain access to otherwise (Conine & Peratoner, 2019; Berggren & Söderlund, 2011). 5.2 Practical and policy implications For universities, the framework offers a template that can be audited: a proposed program can be checked against the five domains for coverage and the four principles for delivery form. Its most demanding implication is institutional: a credible program requires negotiated partnerships with employing governments before launch, because project access and organisational validation are structural components, not enhancements. For city governments, the framework provides a procurement standard: programs that cannot name the live problems participants will work on, or the mechanism by which the organisation validates outcomes, are unlikely to produce the organisational effects purchased. For accreditation and policy bodies, the analysis suggests that the scarcity of organisational-level evidence (Sørensen, 2023) is partly a design failure of past programs, and that requiring multi-level evaluation in program approval would generate the evidence base the field lacks. For both sides, competency instruments already available (Kwon et al., 2025; Tsoutsa et al., 2024) can serve as entry diagnostics and outcome measures, giving programs a common metric. The framework also carries implications for equity between cities. Executive education concentrates where purchasing power concentrates, and the reviewed evidence shows that capacity constraints are most severe precisely where provision is thinnest, in smaller municipalities and developing-country systems (Tan & Taeihagh, 2020). Because the architecture's costs lie mainly in facilitation and partnership rather than in physical plant, consortium models, in which several smaller cities share one cohort and peer sets deliberately cross city boundaries, are a direct extension of the cohort-composition principle and a plausible route to provision at scales single small governments cannot sustain (Holmes, 2012; Bhagavathula et al., 2021). Development agencies and national urban ministries funding smart city programs could treat such consortium-based executive education as eligible infrastructure, on the same footing as the technical systems it teaches officials to govern. 5.3 Limitations Three limitations bound the contribution. First, the framework is a conceptual synthesis: no program built to this design has been evaluated here, and the integration of the two literatures, however disciplined, is the author's analytical construction. Second, the evidence bases are asymmetric: the smart city competency literature is recent and partly expert-opinion based, while the action learning evidence derives mostly from business and general public administration education, so the transfer of its effects to smart city content is inferred rather than demonstrated. Third, the reviewed cases overrepresent European and high-capacity cities; the framework's applicability where basic digital infrastructure and administrative capacity are still developing (Tan & Taeihagh, 2020) requires localisation this article can specify only in principle. 5.4 Future research The priority is implementation research: design-based studies that build programs to the framework, instrument them with the multi-level evaluation it specifies, and report effects at the organisational level where evidence is scarcest (Sørensen, 2023). Comparative work should test whether the five domains hold across institutional contexts, particularly in developing-country city systems. Measurement research should connect competency instruments (Kwon et al., 2025) to program outcomes so that entry diagnostics and effect measures share a construct base. Finally, longitudinal alumni studies would test the framework's professional-field claim, that cohort networks diffuse governance practices between cities, which currently rests on the documented but domain-general strengths of executive cohorts (Holmes, 2012). 6. Conclusion This article asked how universities should develop executive academic programs that support smart city governance and planning. Integrating a smart city literature that specifies what senior urban professionals must be able to do with an executive education literature that specifies how experienced professionals learn, it derived five competency domains and four design principles, and combined them into a program design framework whose organising claim is that the participant's own city is the primary learning environment. The framework converts competency checklists into curricular architecture, gives co-production concrete institutional content, and embeds multi-level evaluation as a structural component. Its status is conceptual, and its test is implementation. As cities continue to acquire technological capability faster than governing capability, the institutions best placed to close that gap are those that can join research, pedagogy, and municipal practice in one design, and the framework offered here is a blueprint for doing so. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics Statement. This study is an integrative review of published literature and involved no human participants, personal data, or animal subjects; ethical approval was therefore not required. Data Availability. No new data were created or analysed in this study. All sources reviewed are publicly available through the references listed. References Almulhim, A. I., & Yigitcanlar, T. (2025). Understanding smart governance of sustainable cities: A review and multidimensional framework. Smart Cities, 8(4), Article 113. https://doi.org/10.3390/smartcities8040113 Bastidas, V., Oti-Sarpong, K., Nochta, T., Wan, L., Tang, J., & Schooling, J. (2024). Leadership of urban digital innovation for public value: A competency framework. IET Smart Cities, 6(3), 237–252. https://doi.org/10.1049/smc2.12063 Berggren, C., & Söderlund, J. (2011). Management education for practicing managers: Combining academic rigor with personal change and organizational action. Journal of Management Education, 35(3), 377–405. https://doi.org/10.1177/1052562910390369 Bhagavathula, S., Brundiers, K., Stauffacher, M., & Kay, B. (2021). Fostering collaboration in city governments' sustainability, emergency management and resilience work through competency-based capacity building. International Journal of Disaster Risk Reduction, 63, Article 102408. https://doi.org/10.1016/j.ijdrr.2021.102408 Conine, T. E., Jr., & Peratoner, C. B. (2019). Applying action learning to corporate projects in executive education programs. Global Business and Organizational Excellence, 38(2), 23–30. https://doi.org/10.1002/joe.21907 David, N., & McNutt, J. (2019). Building a workforce for smart city governance: Challenges and opportunities for the planning and administrative professions. Informatics, 6(4), Article 47. https://doi.org/10.3390/informatics6040047 Dembski, F., Wössner, U., Letzgus, M., Ruddat, M., & Yamu, C. (2020). Urban digital twins for smart cities and citizens: The case study of Herrenberg, Germany. Sustainability, 12(6), Article 2307. https://doi.org/10.3390/su12062307 Gascó-Hernández, M., Nasi, G., Cucciniello, M., & Hiedemann, A. M. (2022). The role of organizational capacity to foster digital transformation in local governments: The case of three European smart cities. Urban Governance, 2(2), 236–246. https://doi.org/10.1016/j.ugj.2022.09.005 Guenduez, A. A., & Mergel, I. (2022). The role of dynamic managerial capabilities and organizational readiness in smart city transformation. Cities, 129, Article 103791. https://doi.org/10.1016/j.cities.2022.103791 Haraguchi, M., Funahashi, T., & Biljecki, F. (2024). Assessing governance implications of city digital twin technology: A maturity model approach. Technological Forecasting and Social Change, 204, Article 123409. https://doi.org/10.1016/j.techfore.2024.123409 Hiedemann, A. M., Nasi, G., & Saporito, R. (2017). A public service-dominant logic for the executive education of public managers. Teaching Public Administration, 35(1), 66–87. https://doi.org/10.1177/0144739416665881 Holmes, M. H. (2012). Raising the ranks of public sector leaders: Results of a national survey of executive Masters of Public Administration programs. Public Personnel Management, 41(3), 449–463. https://doi.org/10.1177/009102601204100304 Iatrellis, O., Panagiotakopoulos, T., Gerogiannis, V. C., Fitsilis, P., & Kameas, A. (2021). Cloud computing and semantic web technologies for ubiquitous management of smart cities-related competences. Education and Information Technologies, 26(2), 2143–2164. https://doi.org/10.1007/s10639-020-10351-9 Kelliher, F., & Byrne, S. (2018). The thinking behind the action (learning): Reflections on the design and delivery of an executive management program. Journal of Work-Applied Management, 10(1), 35–49. https://doi.org/10.1108/JWAM-05-2018-0010 König, P. D. (2021). Citizen-centered data governance in the smart city: From ethics to accountability. Sustainable Cities and Society, 75, Article 103308. https://doi.org/10.1016/j.scs.2021.103308 Kwon, B., Roh, I., & Ro, Y. (2025). Competency checklists and education recommender programs: Training smart city managers. Journal of Urban Technology, 32(2), 107–131. https://doi.org/10.1080/10630732.2025.2475520 Meijer, A., & Rodríguez Bolívar, M. P. (2016). Governing the smart city: A review of the literature on smart urban governance. International Review of Administrative Sciences, 82(2), 392–408. https://doi.org/10.1177/0020852314564308 Mora, L., Gerli, P., Ardito, L., & Messeni Petruzzelli, A. (2023). Smart city governance from an innovation management perspective: Theoretical framing, review of current practices, and future research agenda. Technovation, 123, Article 102717. https://doi.org/10.1016/j.technovation.2023.102717 Paskaleva, K., Evans, J., Martin, C., Linjordet, T., Yang, D., & Karvonen, A. (2017). Data governance in the sustainable smart city. Informatics, 4(4), Article 41. https://doi.org/10.3390/informatics4040041 Perusso, A., van der Sijde, P., Leal, R., & Blankesteijn, M. (2021). The effectiveness and impact of action learning on business graduates' professional practice. Journal of Management Education, 45(2), 177–205. https://doi.org/10.1177/1052562920940374 Rasmussen, K., & Callan, D. (2016). Schools of public policy and executive education: An opportunity missed? Policy and Society, 35(4), 397–411. https://doi.org/10.1016/j.polsoc.2016.12.002 Ruane, M. (2016). Action learning in postgraduate executive management education: An account of practice. Action Learning: Research and Practice, 13(3), 272–280. https://doi.org/10.1080/14767333.2016.1220175 Ruane, M., & Corlett, S. (2024). Participant perspectives on action learning in higher education: A framework for learning. Action Learning: Research and Practice, 21(2), 144–158. https://doi.org/10.1080/14767333.2024.2355912 Ruhlandt, R. W. S. (2018). The governance of smart cities: A systematic literature review. Cities, 81, 1–23. https://doi.org/10.1016/j.cities.2018.02.014 Schrotter, G., & Hürzeler, C. (2020). The digital twin of the city of Zurich for urban planning. PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science, 88(1), 99–112. https://doi.org/10.1007/s41064-020-00092-2 Sørensen, P. (2023). The effects of continuing public sector management education: A systematic literature review. International Journal of Public Sector Management, 36(4/5), 300–314. https://doi.org/10.1108/IJPSM-01-2022-0001 Stephens, S., & Margey, M. (2015). Action learning and executive education: Achieving credible personal, practitioner and organisational learning. Action Learning: Research and Practice, 12(1), 37–51. https://doi.org/10.1080/14767333.2014.993592 Tan, S. Y., & Taeihagh, A. (2020). Smart city governance in developing countries: A systematic literature review. Sustainability, 12(3), Article 899. https://doi.org/10.3390/su12030899 Tomor, Z., Meijer, A., Michels, A., & Geertman, S. (2019). Smart governance for sustainable cities: Findings from a systematic literature review. Journal of Urban Technology, 26(4), 3–27. https://doi.org/10.1080/10630732.2019.1651178 Tsoutsa, P., Panagiotakopoulos, T., Damasiotis, V., & Fitsilis, P. (2024). Empowering urban futures: Key competencies for smart city resilience officers. Urban Governance, 4(3), 180–192. https://doi.org/10.1016/j.ugj.2024.07.001 Viale Pereira, G., Parycek, P., Falco, E., & Kleinhans, R. (2018). Smart governance in the context of smart cities: A literature review. Information Polity, 23(2), 143–162. https://doi.org/10.3233/IP-170067 Weil, C., Bibri, S. E., Longchamp, R., Golay, F., & Alahi, A. (2023). Urban digital twin challenges: A systematic review and perspectives for sustainable smart cities. Sustainable Cities and Society, 99, Article 104862. https://doi.org/10.1016/j.scs.2023.104862 Ziosi, M., Hewitt, B., Juneja, P., Taddeo, M., & Floridi, L. (2024). Smart cities: Reviewing the debate about their ethical implications. AI & Society, 39(3), 1185–1200. https://doi.org/10.1007/s00146-022-01558-0 Hashtags: #SDG11 #SustainableCitiesAndCommunities #SDG4 #QualityEducation #SDG16 #StrongInstitutions #ExecutiveEducation #SmartCityGovernance #UrbanPlanning #PublicSectorCapacity #ActionLearning #CurriculumDesign #U7YJournal #U7Y #AcademicResearch

Author: Daniel Cohen Affiliation: Swiss International University (SIU) ORCID ID: 0009-0000-2881-1030 Submitted 17 March 2026; Revised 22 May 2026; Accepted 11 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10008 Volume 3, December 2026, (SpecSDG10008) Abstract Transnational education separates the place where academic quality is produced from the place where institutional reputation was earned. Degrees are delivered through branch campuses, franchises, and partnerships in one country on the strength of a name built in another, and the governance structures that should hold the two together are strained by distance, regulatory plurality, and information asymmetry. This article develops a conceptual framework, the quality architecture, that specifies how internal quality production, external validation, transparency practices, and the reputational environment must connect for transnational provision to be governed credibly. The framework is built through a structured conceptual analysis of thirty verified peer-reviewed and seminal sources spanning research on transnational education governance, organisational legitimacy, quality assurance and quality culture, and university disclosure. Four components are specified: an internal quality core in which assurance routines and quality culture produce quality; an external validation ring of home, host, and international review; a transparency interface through which governance information reaches stakeholders; and a reputational environment in which rankings and legitimacy judgments circulate. The framework's central claim is that the transparency interface is the load-bearing component in transnational settings: where it is thin, reputation decouples from quality, producing the legitimacy gaps documented in the empirical literature. Four propositions are derived for empirical testing, and implications are drawn for institutions designing cross-border governance, for regulators sequencing oversight, and for research on the reputation-quality relationship in globalised higher education. Keywords: transnational education, institutional reputation, quality assurance, legitimacy, transparency, governance, international branch campuses 1. Introduction Transnational education has made the university mobile. Programmes and providers now move to students rather than the reverse, through international branch campuses, franchised degrees, joint institutes, and distance provision (Knight, 2016). The scale of this movement has turned a once-marginal activity into a structural feature of global higher education, and with it has come a distinctive governance problem: the place where academic quality is produced is no longer the place where the institution's reputation was earned. A student in a host country enrols on the strength of a name, a ranking, and an accreditation earned elsewhere; the teaching, assessment, staffing, and academic culture that will determine their actual education are produced locally, under conditions the home institution only partially controls (Healey, 2015; Lane et al., 2024). The problem is not hypothetical. The sector's history includes host regimes that have closed large numbers of underperforming partnerships, providers that have withdrawn from ventures at reputational and financial cost, and recurring findings that the equivalence promised between home and offshore provision is difficult to deliver in staffing, resources, and student experience (Tran et al., 2023; Hou et al., 2018). At the same time, the reputational instruments that carry cross-border enrolment, rankings and institutional names, are constructed almost entirely from home-campus research performance and sedimented prestige, and contain no information about offshore delivery at all (Selten et al., 2020). The student who chooses a branch campus on the strength of a global ranking is, in informational terms, choosing on evidence about a different campus. The literatures that should illuminate this problem have developed apart. Research on quality assurance in transnational education documents regulatory diversity across host systems, uneven internal assurance, and persistent difficulty in delivering home-campus equivalence abroad (Coleman, 2003; Hou et al., 2018; Mok & Khai, 2024; Tran et al., 2023). Research on institutional reputation and rankings documents how reputational judgments form, circulate, and shape behaviour, largely without reference to cross-border provision (Miotto et al., 2020; Brankovic et al., 2023; Fernández-Gubieda & Gutiérrez-García, 2025). Research on university transparency and disclosure examines what institutions reveal about their governance and performance, almost entirely in domestic settings (Ntim et al., 2017; Ramírez & Tejada, 2018). Each literature holds a piece of the transnational governance problem; none assembles them. There is currently no framework that specifies how quality production, external validation, transparency, and reputation are supposed to connect in transnational provision, and therefore no principled way to diagnose the failures that occur when they do not. The stakes of this gap are practical as well as theoretical. Empirical work shows that students choosing transnational programmes rely heavily on rankings and institutional reputation (Dowling-Hetherington, 2020; Soysal et al., 2024), while the systematic evidence shows that quality delivery in such programmes is the sector's most persistent challenge (Tran et al., 2023). When reputation carries the enrolment decision and quality varies beneath it, the difference is absorbed by students, and eventually by the reputation itself, since reputational damage from failed ventures flows back to the home institution (Healey, 2015; Wilkins, 2017). This article addresses the gap with a conceptual contribution. It asks three questions. First, what does the existing evidence establish about how quality is produced, validated, and communicated in transnational education? Second, how do reputation and legitimacy operate as governing forces on transnational providers? Third, how can these elements be integrated into a framework that both explains observed failures and guides transparent governance design? The answer is the quality architecture: a four-component framework, developed in Section 4, whose central claim is that transparency is the load-bearing connection between quality production and reputational judgment in cross-border settings. Section 2 reviews the contributing literatures, Section 3 describes the method, Section 5 discusses implications and limitations, and Section 6 concludes. 2. Literature Review 2.1 The governance problem of transnational education Definitional work distinguishes collaborative transnational provision, delivered with local partners, from independent provision such as wholly owned branch campuses, and shows that governance obligations differ across the resulting forms (Knight, 2016). Systematic mapping of the field finds quality assurance the most researched topic, alongside policy, cultural difference, and student experience, but also finds the literature fragmented and often thin on comparative theory (Kosmützky & Putty, 2016). The empirical record documents the governance problem from several directions. Early analysis of branch campuses in Southeast Asia described them as among the most intrusive yet least monitored forms of cross-border provision and traced the first attempts of home-country agencies to extend oversight abroad (Coleman, 2003). Comparative study of host regulation in Malaysia, Singapore, China, and South Korea finds external quality assurance approaches ranging from exemption through reliance on home accreditation to full duplication, with regulatory liberalism and comprehensiveness varying by country (Hou et al., 2018). Reviews of Asian transnationalisation conclude that quality assurance practice is primarily shaped by host-country and provider policies rather than by any common standard (Mok & Khai, 2024). Host-country research in Ghana documents benefits in curriculum and pedagogy alongside challenges of unclear policy guidelines, resourcing, and curricular localisation (Owusu-Agyeman & Amoakohene, 2020). A systematic review across two decades of research identifies quality assurance, equivalence of student experience, staff preparedness, and local adaptation as the sector's central challenges, and a proper regulatory framework and global data collection as success conditions (Tran et al., 2023). Two theoretical treatments sharpen the problem. A risk-based typology reframes transnational forms by the degree of reputational and operational risk they transfer to the home institution, arguing that activity-based classifications obscure what governance must actually manage (Healey, 2015). Principal-agent analysis of branch campuses in Malaysia, Qatar, and the United Arab Emirates models the home institution, host state, and campus as a governance triangle in which information asymmetry between principals and agents is the fundamental condition to be managed (Lane et al., 2024). Both treatments converge on the same insight: transnational governance is, at core, a problem of credible information across distance. 2.2 Reputation, rankings, and legitimacy The second literature explains the forces that carry enrolment and standing. Legitimacy theory distinguishes pragmatic, moral, and cognitive legitimacy and analyses the strategies by which organisations gain, maintain, and repair each (Suchman, 1995). Applied to higher education, survey evidence shows that reputation exerts a significant positive effect on legitimacy, and that both function as intangible assets underpinning competitive advantage (Miotto et al., 2020). Institutional theory explains branch campus establishment itself partly as legitimacy- and status-seeking behaviour (Wilkins & Huisman, 2012), and case research on Chinese branch campuses in Southeast Asia shows providers actively constructing legitimacy through conformity, selective conformity, and creation strategies according to local dependence (He & Wilkins, 2018). Strategic-plan analysis across seventy-eight universities in thirty-three countries finds legitimacy-seeking stratified by status: high-ranked institutions emphasise global positioning while others anchor claims in local relevance (Stensaker et al., 2019). Rankings are the most consequential reputational instrument. Longitudinal analysis shows the major global rankings reduce largely to two factors, research performance and reputation, and are stable over time (Selten et al., 2020); critical work shows ranking organisations managing their own legitimacy through principles that are decoupled from practice (Barron, 2017), and documents how rankings multiply and reshape reputational geographies rather than merely record quality (Collins & Park, 2016). The institutionalisation of rankings has proceeded through continuities, interdependencies, and the engagement of universities themselves (Brankovic et al., 2023). Crucially for transnational provision, evidence shows international student sorting is driven more by sedimented reputation than by year-to-year ranking movement (Soysal et al., 2024), and branch campus enrolment decisions lean on home-institution rankings and accreditations (Dowling-Hetherington, 2020), exposing providers to reputational risk they cannot fully control (Healey, 2015; Wilkins, 2017). Systematic reviews of university reputation conclude the construct remains fragmented and under-theorised relative to its strategic weight (Amado Mateus & Juarez Acosta, 2022; Fernández-Gubieda & Gutiérrez-García, 2025). 2.3 Quality assurance and quality culture The third literature concerns how quality is actually produced inside institutions. Evidence from accredited universities shows accreditation affecting performance substantially through the mediation of quality culture, the shared commitment that converts procedures into practice (Iqbal et al., 2024). Recent structural modelling finds internal and external assurance exerting direct effects on performance while quality culture functions as an enabling condition shaped by assurance routines (Ulkhaq et al., 2026). For transnational settings, prospective models argue that quality management must be aligned with student expectations and experience from the design stage rather than audited retrospectively (Tsiligiris & Hill, 2021). The implication that matters for the present framework is that quality is produced by the interaction of routines and culture at the point of delivery, which in transnational provision lies at the far end of the governance chain from the reputation that recruits students, and that assurance conducted at distance measures the routines more easily than the culture (Hou et al., 2018; Tran et al., 2023). 2.4 Transparency and disclosure The final literature examines how governance information reaches stakeholders. Analysis of UK universities finds voluntary disclosure highly variable and overall low, with audit committee quality, board diversity, and governance committees associated with fuller disclosure, and concludes that accountability motives are weaker than legitimacy and resource motives (Ntim et al., 2017). Stakeholder research in Spain finds strong demand for disclosure of governance structures and processes as a condition of trust (Ramírez & Tejada, 2018). Studies of highly ranked universities link board structure and stakeholder participation to social-responsibility disclosure (Garde Sánchez et al., 2020), and evidence from Asian universities shows size and complexity predicting sustainability disclosure, interpreted through legitimacy theory (Raimo et al., 2025). Across this literature, disclosure functions simultaneously as accountability and as legitimacy management, and the balance between the two is an empirical variable rather than a given. What the disclosure literature has not yet examined is the transnational case, and the omission matters more than a simple gap in coverage. Every study cited above analyses institutions disclosing to audiences within a single accountability jurisdiction: the funders, regulators, and publics of one country. Transnational provision splits the audience. The stakeholders with the greatest informational need, students, employers, and regulators in host countries, are precisely those least served by home-country reporting conventions, and no reviewed study measures what such stakeholders can learn about an offshore operation from public sources. This unmeasured quantity is, on the argument developed below, the variable on which credible transnational governance turns. Read together, the four literatures supply the components of an unassembled system: a governance problem defined by information asymmetry across borders, a reputational environment that carries decisions, a quality production process rooted in local routines and culture, and a disclosure practice that could connect them but is documented mainly as domestic and discretionary. The framework developed below assembles them. 3. Method 3.1 Design and rationale The study is a conceptual paper built on a structured evidence base. The research questions require theory construction, specifying how constructs from separate literatures relate, rather than the aggregation of comparable empirical results, which rules out systematic review as the primary design; at the same time, the framework's claims are intended to be anchored in verifiable findings, which rules out unstructured essay. The method therefore combines explicit source selection with a disciplined synthesis procedure, and states the framework's conceptual status openly. 3.2 Source identification and criteria Sources were identified between June and August 2026 through structured queries of scholarly databases (Consensus, drawing on Semantic Scholar, Scopus, and related indexes), organised in four query families corresponding to the constructs in the title: transnational education governance and quality assurance; institutional reputation, rankings, and legitimacy; internal quality assurance and quality culture; and university transparency and disclosure. Inclusion criteria were: peer-reviewed publication or seminal theoretical standing; direct relevance to at least one construct; and confirmable bibliographic identity, with every source verified against its digital object identifier or publisher record before use. Exclusion criteria were: promotional literature, sources whose metadata could not be confirmed, and duplicative preprints of published work. Thirty sources satisfied all criteria; one earlier theoretical work (Suchman, 1995) was retained deliberately as the foundational statement of legitimacy theory on which the higher-education applications rest. 3.3 Analytical procedure The synthesis proceeded in three steps. First, each literature was analysed for the function it assigns to its central construct: what quality assurance does, what reputation does, what disclosure does, and for whom. Second, a dependency analysis asked what each function presupposes from the others in a specifically transnational setting, where production, validation, and judgment are geographically and institutionally separated; this generated the four-component structure and identified the transparency interface as the component on which cross-border alignment depends. Third, the draft framework was tested against the empirical record: documented failure patterns in transnational provision were checked against the framework's predictions, and four propositions were formulated where the framework makes claims the existing evidence can test but has not. Throughout, claims are attributed at the strength the underlying studies warrant, and framework-level statements are marked as conceptual. 3.4 Rigor and boundaries Bias was limited by drawing on critical as well as managerial scholarship in each literature, by verifying every source against the original record, and by requiring that each framework component rest on multiple independent sources. The analysis is bounded in three ways: it synthesises published research rather than new fieldwork; the transnational evidence base is concentrated on branch campuses and Asian host systems, reflecting the field's own concentration (Kosmützky & Putty, 2016); and the framework is a set of theoretical claims whose empirical adequacy the propositions are designed to expose to testing. These boundaries are revisited in the Limitations subsection. 4. The Quality Architecture Framework The framework models transparent governance in transnational education as an architecture of four connected components. Figure 1 presents the structure, and Table 1 specifies each component's function, mechanisms, and characteristic transnational risks. The subsections develop the components and then the alignment dynamics that constitute the framework's explanatory core. 4.1 The internal quality core At the centre sits the process that actually produces educational quality: curriculum, teaching, assessment, staffing, and the assurance routines and quality culture in which they are embedded. The evidence is consistent that routines and culture interact, with culture converting procedure into practice (Iqbal et al., 2024) and assurance mechanisms shaping culture over time (Ulkhaq et al., 2026). In transnational provision this core is duplicated at distance: branch campuses and partners must reproduce not only the home institution's procedures but the practice-sustaining culture beneath them, under different staffing profiles and resource conditions, which the sector evidence identifies as its hardest problem (Tran et al., 2023; Hou et al., 2018). The core's characteristic transnational failure is hollow replication: procedures formally transplanted, culture not. Why hollow replication is the default rather than the exception follows from the sector's operating conditions. Offshore operations rely more heavily on visiting, adjunct, and locally recruited staff whose induction into the home institution's academic culture is brief or absent; resource levels are typically thinner than at the home campus; and the practices that carry quality culture, moderation conversations, examiner networks, informal collegial scrutiny, do not travel in policy documents (Tran et al., 2023; Owusu-Agyeman & Amoakohene, 2020). Prospective quality models respond to exactly this condition by requiring alignment between quality management and the actual expectations and experience of offshore students from the design stage, rather than treating equivalence as something to be audited after the fact (Tsiligiris & Hill, 2021). The framework adopts that prospective logic: the core is a system to be built deliberately in each location, not an inheritance that arrives with the brand. 4.2 The external validation ring Around the core operates the ring of external validators: home-country agencies extending oversight abroad, host-country regulators, and international accreditors. The comparative record shows this ring is plural and uneven, spanning exemption, reliance on home accreditation, duplication, and international review (Hou et al., 2018), primarily shaped by host and provider policy rather than common standards (Mok & Khai, 2024), and historically lagging the expansion of provision itself (Coleman, 2003). Principal-agent analysis explains the ring's structural weakness: each validator holds partial information about an agent whose principals sit in different jurisdictions, so validation fragments precisely where provision crosses borders (Lane et al., 2024). The ring's characteristic failure is the gap between overlapping mandates, in which no validator observes the whole operation. 4.3 The transparency interface The third component is the framework's distinctive contribution: the set of practices through which the institution makes its quality production and validation visible to stakeholders, spanning governance disclosure, outcome reporting, and honest representation of the relationship between home and offshore provision. The domestic evidence shows such disclosure is discretionary, variable, and governance-dependent (Ntim et al., 2017; Garde Sánchez et al., 2020), demanded by stakeholders as a condition of trust (Ramírez & Tejada, 2018), and used by institutions as legitimacy management as much as accountability (Raimo et al., 2025). The framework's claim, developed from the dependency analysis, is that in transnational settings this interface is load-bearing: because students, employers, and regulators in the host country cannot observe the internal core directly and the validation ring is fragmented, disclosed information is the principal channel through which quality can discipline reputation. Where the interface is thin, reputational judgment forms on rankings and name alone, and the enrolment evidence shows it does (Dowling-Hetherington, 2020; Soysal et al., 2024). This claim is conceptual, but it organises the observed pattern: the sector's documented failures concentrate where disclosure about offshore operations is weakest (Healey, 2015; Wilkins, 2017). 4.4 The reputational environment The outer component is the environment in which reputational and legitimacy judgments circulate: rankings and their institutionalised infrastructure (Brankovic et al., 2023; Selten et al., 2020), stakeholder perception aggregated over time (Fernández-Gubieda & Gutiérrez-García, 2025), and the legitimacy types through which audiences grant standing (Suchman, 1995; Miotto et al., 2020). Transnational providers both draw on this environment, borrowing home reputation to enter host markets (Wilkins & Huisman, 2012; He & Wilkins, 2018), and are exposed to it, since offshore failure flows back through the same channels (Healey, 2015). The environment's characteristic property is inertia: reputation is sedimented and moves slowly relative to quality (Soysal et al., 2024; Collins & Park, 2016), which is precisely what makes decoupling possible. 4.5 Alignment, decoupling, and four propositions The framework's explanatory core is the relationship among components. When the four are aligned, quality produced in the core is validated by the ring, made visible through the interface, and reflected, with lag, in the environment; reputation then approximates a trailing indicator of quality. Decoupling occurs when the interface fails: quality and reputation cease to constrain one another, hollow replication in the core persists unobserved, and the legitimacy eventually granted or withdrawn is disconnected from educational reality. The framework distinguishes two decoupling paths with different dynamics. Upward decoupling occurs when borrowed reputation exceeds delivered quality: enrolment and fees flow on the strength of the home name while the offshore core underperforms, a configuration that is privately stable for the provider in the short run and is therefore the one regulation must actively disturb. Downward decoupling occurs when delivered quality exceeds recognised reputation: a competent offshore operation is discounted because host stakeholders cannot verify what it does, a configuration that penalises exactly the providers investing most in the core. Both paths are failures of the same interface, which is why the framework treats disclosure as the single point of intervention that acts on both: richer disclosure erodes the private stability of upward decoupling and supplies the verification that resolves downward decoupling. The legitimacy literature adds a temporal asymmetry: legitimacy is slow to build, faster to lose, and hardest to repair once audiences reclassify a provider from competent to untrustworthy (Suchman, 1995), which converts sustained upward decoupling into a stock of latent reputational liability rather than a maintainable equilibrium (He & Wilkins, 2018; Healey, 2015). Four propositions state the framework's testable content. First, transnational operations with richer public disclosure of offshore governance and outcomes will show smaller home-host differences in educational quality, because visibility disciplines replication. Second, the effect of external validation on offshore quality will be mediated by the local quality culture it induces rather than by compliance alone, extending the domestic mediation evidence (Iqbal et al., 2024; Ulkhaq et al., 2026) to cross-border settings. Third, reputational damage from offshore failure will be greater, and repair slower, for institutions whose prior disclosure was thin, because thin disclosure converts failure from a performance problem into a trust problem (Suchman, 1995; Ntim et al., 2017). Fourth, host-country regulatory regimes that mandate disclosure will produce convergence between reputation and quality faster than regimes that mandate validation alone (Hou et al., 2018; Lane et al., 2024). Table 1 Components of the Quality Architecture: Functions, Mechanisms, and Transnational Risks Component Function Principal mechanisms Characteristic transnational risk Key evidence Internal quality core Produces educational quality Assurance routines; quality culture; staffing and assessment practice Hollow replication: procedures transplanted without culture Iqbal et al. (2024); Ulkhaq et al. (2026); Tran et al. (2023) External validation ring Certifies quality to outside audiences Home-agency oversight; host regulation; international accreditation Fragmented mandates; no validator sees the whole operation Coleman (2003); Hou et al. (2018); Lane et al. (2024) Transparency interface Makes production and validation visible Governance disclosure; outcome reporting; honest representation of offshore provision Thin, discretionary disclosure; legitimacy management displacing accountability Ntim et al. (2017); Ramírez & Tejada (2018); Raimo et al. (2025) Reputational environment Allocates standing and enrolment Rankings; sedimented reputation; legitimacy judgments Inertia enabling decoupling; borrowed reputation outrunning local quality Selten et al. (2020); Soysal et al. (2024); Wilkins & Huisman (2012) Note. Components are the author's synthesis of the reviewed literatures; the transnational risks column states the failure mode the framework predicts when the component operates without the support of the others. Full argument in Sections 4.1–4.5. Figure 1. The quality architecture for transparent governance in transnational education. Quality is produced in the internal core, certified by the external validation ring, and made visible through the transparency interface to the reputational environment; the interface is the load-bearing connection, and its failure decouples reputation from quality. Source: author's elaboration based on the reviewed literature. 5. Discussion 5.1 Theoretical implications The framework contributes to three conversations. To transnational education research, it supplies the integrating structure the field's own reviews call for (Kosmützky & Putty, 2016; Tran et al., 2023): quality assurance, risk typologies, and legitimacy strategies cease to be separate topics and become components whose interaction explains outcomes. The reframing also changes what counts as a finding. Studies that document regulatory diversity (Hou et al., 2018) or student reliance on rankings (Dowling-Hetherington, 2020) currently read as descriptions of separate phenomena; within the architecture they become measurements of particular components whose joint configuration, not their individual states, predicts whether provision is credibly governed. The principal-agent account of governance risk (Lane et al., 2024) is extended by specifying disclosure as the institutional answer to the information asymmetry that theory identifies: transparency is what converts an unobservable agent into a governable one. To legitimacy theory, the framework adds a boundary case with general interest: transnational provision separates the audience that grants legitimacy from the activity that should earn it, making the management of legitimacy through disclosure empirically distinguishable from the management of quality itself, a distinction that domestic settings blur (Suchman, 1995; Raimo et al., 2025). To reputation research, which its own reviews describe as fragmented (Fernández-Gubieda & Gutiérrez-García, 2025), the framework offers a mechanism linking reputation to its supposed referent: reputation tracks quality only through the information that crosses the interface, which explains both the documented inertia of reputational judgment (Soysal et al., 2024) and the possibility of its decoupling. The framework also speaks to an older question in institutional theory: when disclosure serves legitimacy management, does it inform or merely reassure (Ntim et al., 2017; Raimo et al., 2025)? The transnational setting sharpens the question because the audiences are segmented. Disclosure aimed at home-country audiences, annual reports, sustainability statements, strategic plans, can legitimate without informing host-country stakeholders at all, since the information those stakeholders need concerns the offshore operation specifically. The framework therefore implies that the informativeness of disclosure must be assessed audience by audience, and predicts that providers under reputational pressure will disclose most where it legitimates and least where it informs, a prediction consistent with the domestic finding that accountability motives run weaker than legitimacy motives (Ntim et al., 2017). 5.2 Practical and policy implications For institutions, the framework functions as a governance audit. A transnational operation can be examined component by component: whether quality culture, not merely procedure, has been reproduced offshore; whether any single validator observes the whole operation; what a host-country stakeholder can actually learn about the offshore operation from public sources; and how exposed the home reputation is to offshore failure. The framework implies that the cheapest robust investment is usually the interface: disclosure of offshore governance, staffing, and outcomes is within institutional control, unlike host regulation or ranking behaviour, and on the framework's first proposition it disciplines the core as a by-product. For regulators, the fourth proposition implies a sequencing: mandating disclosure may achieve more, sooner, than adding validation layers to an already fragmented ring (Hou et al., 2018). For students and their advisers, the framework identifies the questions that rankings do not answer and disclosure should: who teaches, who examines, and which quality regime actually governs the campus at which they will study (Dowling-Hetherington, 2020; Tsiligiris & Hill, 2021). The framework further suggests a minimum disclosure standard that institutions and quality agencies could adopt without new regulation. For each offshore operation it would cover the governance relationship between home and host entities; the proportion and status of home-appointed academic staff; which examinations and awards are subject to home-campus moderation; which external validator holds current oversight and when it last reviewed the operation; and comparable outcome indicators for home and offshore cohorts. Each element corresponds to a documented failure locus in the empirical record (Tran et al., 2023; Hou et al., 2018; Lane et al., 2024), and none exceeds what institutions already hold internally. Publishing such a standard would also give accreditors and ranking organisations a defensible way to incorporate transnational provision into instruments that currently ignore it, addressing a measurement blind spot the rankings literature has identified from within (Barron, 2017; Selten et al., 2020). 5.3 Limitations Four limitations bound these claims. First, the framework is conceptual: its components are grounded in verified empirical literature, but their integration, and especially the load-bearing status of the transparency interface, is the author's theoretical construction, stated as propositions precisely because it awaits testing. Second, the evidence base inherits the field's concentration on branch campuses and Asian host systems; collaborative and franchise provision, where partner incentives differ, may stress the architecture differently. Third, the disclosure literature the interface rests on is domestic; transnational disclosure may face constraints, including host-country political sensitivity, that domestic studies do not capture (Wilkins, 2017). Fourth, the framework treats the reputational environment as exogenous inertia, whereas ranking organisations are themselves strategic actors (Barron, 2017); endogenising their behaviour is beyond this article's scope. 5.4 Future research The propositions define the agenda. Testing the first requires paired measurement of disclosure richness and home-host quality differences across transnational operations, feasible through document analysis combined with the quality data that some host regulators already collect (Hou et al., 2018). The second extends established mediation designs (Iqbal et al., 2024; Ulkhaq et al., 2026) to offshore campuses. The third invites event-study methods around documented offshore failures, relating reputational impact to prior disclosure. The fourth calls for comparative regulatory analysis across host systems with different disclosure mandates. Beyond the propositions, qualitative work on how offshore staff experience the tension between representing quality and producing it would illuminate the interface from inside, and extension of the framework to collaborative provision and to emerging host regions would test its generality (Owusu-Agyeman & Amoakohene, 2020; Kosmützky & Putty, 2016). 6. Conclusion This article set out to connect what research on transnational education has kept apart: the production of quality, its validation, its communication, and the reputational judgments that carry cross-border enrolment. The quality architecture framework specifies four components and locates the decisive connection in the transparency interface, through which alone quality can discipline reputation across borders. The framework explains the sector's documented failure patterns as decoupling phenomena, yields four testable propositions, and converts abstract calls for accountability into a specific design claim: in transnational education, transparent governance is not a virtue appended to quality management but the structural condition under which reputation and quality remain connected at all. As provision continues to globalise, the institutions and regulators that treat disclosure as infrastructure, rather than as discretionary communication, will be the ones whose reputations remain worth borrowing. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics Statement. This study is a conceptual analysis of published literature and involved no human participants, personal data, or animal subjects; ethical approval was therefore not required. Data Availability. No new data were created or analysed in this study. All sources reviewed are publicly available through the references listed. References Amado Mateus, M., & Juarez Acosta, F. (2022). Reputation in higher education: A systematic review. Frontiers in Education, 7, Article 925117. https://doi.org/10.3389/feduc.2022.925117 Barron, G. R. S. (2017). The Berlin Principles on Ranking Higher Education Institutions: Limitations, legitimacy, and value conflict. Higher Education, 73(2), 317–333. https://doi.org/10.1007/s10734-016-0022-z Brankovic, J., Hamann, J., & Ringel, L. (2023). The institutionalization of rankings in higher education: Continuities, interdependencies, engagement. Higher Education, 86(4), 719–731. https://doi.org/10.1007/s10734-023-01018-8 Coleman, D. (2003). Quality assurance in transnational education. Journal of Studies in International Education, 7(4), 354–378. https://doi.org/10.1177/1028315303255597 Collins, F. L., & Park, G.-S. (2016). Ranking and the multiplication of reputation: Reflections from the frontier of globalizing higher education. Higher Education, 72(1), 115–129. https://doi.org/10.1007/s10734-015-9941-3 Dowling-Hetherington, L. (2020). Transnational higher education and the factors influencing student decision-making: The experience of an Irish university. Journal of Studies in International Education, 24(3), 291–313. https://doi.org/10.1177/1028315319826320 Fernández-Gubieda, S., & Gutiérrez-García, E. (2025). University reputation overlooked: A systematic literature review of an under-researched concept. Corporate Reputation Review. Advance online publication. https://doi.org/10.1057/s41299-025-00234-5 Garde Sánchez, R., Flórez-Parra, J. M., López-Pérez, M. V., & López-Hernández, A. M. (2020). Corporate governance and disclosure of information on corporate social responsibility: An analysis of the top 200 universities in the Shanghai Ranking. Sustainability, 12(4), Article 1549. https://doi.org/10.3390/su12041549 He, L., & Wilkins, S. (2018). Achieving legitimacy in cross-border higher education: Institutional influences on Chinese international branch campuses in South East Asia. Journal of Studies in International Education, 22(3), 179–197. https://doi.org/10.1177/1028315317738774 Healey, N. M. (2015). Towards a risk-based typology for transnational education. Higher Education, 69(1), 1–18. https://doi.org/10.1007/s10734-014-9757-6 Hou, A. Y. C., Hill, C., Chen, K. H. J., & Tsai, S. (2018). A comparative study of international branch campuses in Malaysia, Singapore, China, and South Korea: Regulation, governance, and quality assurance. Asia Pacific Education Review, 19(4), 543–555. https://doi.org/10.1007/s12564-018-9550-9 Iqbal, S., Taib, C. A. B., & Razalli, M. R. (2024). The effect of accreditation on higher education performance through quality culture mediation: The perceptions of administrative and quality managers. The TQM Journal, 36(2), 572–592. https://doi.org/10.1108/TQM-11-2022-0322 Knight, J. (2016). Transnational education remodeled: Toward a common TNE framework and definitions. Journal of Studies in International Education, 20(1), 34–47. https://doi.org/10.1177/1028315315602927 Kosmützky, A., & Putty, R. (2016). Transcending borders and traversing boundaries: A systematic review of the literature on transnational, offshore, cross-border, and borderless higher education. Journal of Studies in International Education, 20(1), 8–33. https://doi.org/10.1177/1028315315604719 Lane, J. E., Borgos, J., & Schueller, J. (2024). Analysing risk in transnational higher education governance: A principal-agent analysis of international branch campuses. Compare: A Journal of Comparative and International Education. Advance online publication. https://doi.org/10.1080/03057925.2024.2374300 Miotto, G., Del-Castillo-Feito, C., & Blanco-González, A. (2020). Reputation and legitimacy: Key factors for Higher Education Institutions' sustained competitive advantage. Journal of Business Research, 112, 342–353. https://doi.org/10.1016/j.jbusres.2019.11.076 Mok, K. H., & Khai, T. S. (2024). Transnationalization of higher education in China and Asia: Quality assurance and students' learning experiences. Asian Education and Development Studies, 13(3), 208–226. https://doi.org/10.1108/AEDS-01-2024-0004 Ntim, C. G., Soobaroyen, T., & Broad, M. J. (2017). Governance structures, voluntary disclosures and public accountability: The case of UK higher education institutions. Accounting, Auditing & Accountability Journal, 30(1), 65–118. https://doi.org/10.1108/AAAJ-10-2014-1842 Owusu-Agyeman, Y., & Amoakohene, G. (2020). Transnational education delivery in Ghana: Examining the benefits, challenges and future prospects. Policy Reviews in Higher Education, 4(2), 135–163. https://doi.org/10.1080/23322969.2020.1774408 Raimo, N., L'Abate, V., Albergo, F., & Vitolla, F. (2025). From commitment to transparency: What drives SDG disclosure in Asian universities? Sustainable Development, 33(5), 7546–7556. https://doi.org/10.1002/sd.3541 Ramírez, Y., & Tejada, Á. (2018). Corporate governance of universities: Improving transparency and accountability. International Journal of Disclosure and Governance, 15(1), 29–39. https://doi.org/10.1057/s41310-018-0034-2 Selten, F., Neylon, C., Huang, C.-K., & Groth, P. (2020). A longitudinal analysis of university rankings. Quantitative Science Studies, 1(3), 1109–1135. https://doi.org/10.1162/qss_a_00052 Soysal, Y. N., Baltaru, R.-D., & Cebolla-Boado, H. (2024). Meritocracy or reputation? The role of rankings in the sorting of international students across universities. Globalisation, Societies and Education, 22(2), 252–263. https://doi.org/10.1080/14767724.2022.2070131 Stensaker, B., Lee, J. J., Rhoades, G., Ghosh, S., Castiello-Gutiérrez, S., Vance, H., Çalıkoğlu, A., Kramer, V., Liu, S., Marei, M. S., O'Toole, L., Pavlyutkin, I., & Peel, C. (2019). Stratified university strategies: The shaping of institutional legitimacy in a global perspective. The Journal of Higher Education, 90(4), 539–562. https://doi.org/10.1080/00221546.2018.1513306 Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20(3), 571–610. https://doi.org/10.5465/amr.1995.9508080331 Tran, N. H. N., Amado, C. A. d. E. F., & Santos, S. P. d. (2023). Challenges and success factors of transnational higher education: A systematic review. Studies in Higher Education, 48(1), 113–136. https://doi.org/10.1080/03075079.2022.2121813 Tsiligiris, V., & Hill, C. (2021). A prospective model for aligning educational quality and student experience in international higher education. Studies in Higher Education, 46(2), 228–244. https://doi.org/10.1080/03075079.2019.1628203 Ulkhaq, M. M., Handayani, N. U., & Rahmat, T. O. (2026). Do quality assurance mechanisms work through quality culture or directly? Evidence on higher education performance from Indonesia. Journal of Applied Research in Higher Education. Advance online publication. https://doi.org/10.1108/JARHE-09-2025-0808 Wilkins, S. (2017). Ethical issues in transnational higher education: The case of international branch campuses. Studies in Higher Education, 42(8), 1385–1400. https://doi.org/10.1080/03075079.2015.1099624 Wilkins, S., & Huisman, J. (2012). The international branch campus as transnational strategy in higher education. Higher Education, 64(5), 627–645. https://doi.org/10.1007/s10734-012-9516-5 Hashtags : #SDG4 #QualityEducation #SDG16 #StrongInstitutions #TransnationalEducation #InstitutionalReputation #QualityAssurance #TransparentGovernance #Legitimacy #InternationalBranchCampuses #HigherEducation #U7YJournal #U7Y #AcademicResearch

Author: Hiroshi Tanaka Affiliation: Swiss International University (SIU) ORCID ID: 0009-0005-5461-8396 Submitted 25 March 2026; Revised 25 May 2026; Accepted 07 July 2026; Available online 08 August 2026; Version of Record 08 August 2026. Doi: https://doi.org/10.65326/u7y.SpecSDG10007 Volume 3, December 2026, (SpecSDG10007) Abstract Vocational education and training (VET) certificates remain the dominant currency through which mid-skill competence is communicated to employers, yet the mechanisms for attesting them were designed for paper documents, national labour markets, and manual verification. This article examines how digital attestation, understood as the cryptographically verifiable issuance, holding, and checking of credentials, can reposition VET certificates within algorithmically mediated job markets and increasingly cross-border workforces. The study is an integrative review of peer-reviewed research and institutional policy sources published mainly between 2015 and 2026, identified through scholarly databases and verified against original records. The analysis is organised through a four-layer conceptual framework developed in the article: a cryptographic trust layer that secures the authenticity of the certificate; a semantic interoperability layer that renders its meaning machine-readable across systems; an institutional recognition layer that connects verification to quality assurance and qualifications frameworks; and a labour-market utilisation layer in which verified credentials function as hiring signals inside data-driven matching platforms. The review finds that technical verification is the most mature layer, while semantic and institutional alignment lag behind and constrain the value of attested certificates for cross-border employment. Equity emerges as a cross-cutting condition, since infrastructural and skills gaps can exclude the very workers VET serves. The framework offers policymakers and credentialing bodies a structured basis for sequencing investment, and identifies the coupling between attestation infrastructure and skills-based hiring as a priority for empirical research. Keywords: digital credentials, vocational education and training, blockchain attestation, micro-credentials, labour mobility, skills recognition, verifiable credentials 1. Introduction The certificate is the point at which education systems and labour markets meet. For vocational education and training, whose graduates typically enter employment directly rather than progressing to further study, the credibility of that document carries particular weight: it is often the principal evidence of competence a worker can present to an employer who has never met them. Yet the infrastructure through which vocational certificates are issued, authenticated, and interpreted has changed far less than the labour markets in which they are used. Employers increasingly recruit through digital platforms that parse applications algorithmically (Broecke, 2023), advertise roles in terms of discrete skills rather than whole qualifications (Bone et al., 2025), and hire across borders at a scale that manual document checking was never designed to support (Chakroun & Keevy, 2018). The mismatch produces two well-documented failures. The first is fraud. Paper certificates and scanned copies are vulnerable to forgery, and verification through direct correspondence with issuing institutions is slow, costly, and inconsistent (Rustemi et al., 2023; Said et al., 2025). Fraudulent credentials corrode the meritocratic function of both education systems and hiring processes, because employers who cannot distinguish genuine from counterfeit documents rationally discount all of them. The second failure is illegibility. Even an authentic vocational certificate frequently fails to communicate, in terms a foreign employer or an algorithmic matching system can interpret, what its holder can actually do (Keevy et al., 2026). A welding qualification from one country reaches a recruiter in another as an unfamiliar document title, not as a structured statement of competences aligned to a recognised framework. Digital attestation addresses both failures in principle. Cryptographic techniques can make certificates tamper-evident and instantly verifiable without contacting the issuer (Capece et al., 2020; Rani et al., 2024). Data standards and qualifications frameworks can make their content machine-readable and comparable across borders (Council of the European Union, 2022). Self-sovereign identity architectures can place the credential under the control of the worker rather than the institution (Grech et al., 2021; Chan et al., 2025). A substantial technical literature now demonstrates feasibility, and policy instruments such as the European digital credentials infrastructure indicate institutional appetite. What the literature has not yet resolved is how these strands connect: research on verification systems, on micro-credentials and recognition policy, and on labour-market signaling has developed in largely separate communities, and studies of blockchain certificate systems rarely engage the economics of hiring, while signaling research rarely considers the verification infrastructure its instruments presuppose. There is, in consequence, no integrated account of what digital attestation must accomplish, across technical, semantic, institutional, and labour-market dimensions at once, for VET certificates to function in smart job markets and to travel with mobile workers. The stakes of the gap are raised by two converging developments that give this article its subtitle. The first is the emergence of what can be called smart job markets: hiring environments in which vacancy data, applicant records, and matching decisions are expressed as structured skill information and processed by algorithmic intermediaries (Colombo et al., 2019; Broecke, 2023). In such markets a credential participates in hiring only to the extent that its content can be verified and parsed without human mediation. The second is the normalisation of cross-border and remote work, which detaches the labour market a vocational graduate can access from the national system that certified them. Both developments reward exactly what conventional attestation cannot deliver: instant verification and machine-readable meaning at negligible marginal cost. This article addresses that gap. It asks two questions. First, what does the existing evidence, technical, educational, and economic, establish about each dimension of digital attestation for vocational certificates? Second, how can these dimensions be integrated into a single framework that explains where current initiatives succeed, where they stall, and what sequencing of effort follows? The contribution is conceptual: an integrative review organised through a four-layer framework of digital attestation, developed inductively from the reviewed literature. The article proceeds as follows. Section 2 reviews the literature thematically. Section 3 describes the review method. Section 4 presents the framework and its four layers. Section 5 discusses implications, limitations, and research directions, and Section 6 concludes. 2. Literature Review 2.1 Certificate fraud and the verification burden The scale of credential fraud is difficult to measure directly, but the research response to it is extensive. A systematic review of blockchain-based academic certificate verification identified thirty-four relevant studies published between 2018 and 2022 from an initial pool of more than seventeen hundred records, and organised them into six themes ranging from issuance architectures to revocation (Rustemi et al., 2023). The consistent motivation across this body of work is that conventional verification, in which an employer or admissions office writes to the issuing institution and awaits confirmation, is slow, expensive, and unreliable, particularly across borders and in systems where records are fragmented or have been destroyed (Khan et al., 2021; Castro & Au-Yong-Oliveira, 2021). Proposed systems anchor certificate hashes or full credential objects on distributed ledgers so that any verifier can confirm authenticity without contacting the issuer. Implementations span permissioned architectures for national education authorities (Khan et al., 2021; Rani et al., 2024), public-chain deployments with decentralised storage designed to cover multiple education levels within one country (Said et al., 2025), and institutional pilots built on open standards such as Blockcerts (Capece et al., 2020). Evaluations report acceptable throughput and latency for realistic transaction volumes (Rani et al., 2024), and comparative analyses find cost and speed advantages over both manual verification and centralised digital registries (Said et al., 2025). Two qualifications recur. First, most systems are prototypes or single-institution pilots; the literature demonstrates feasibility more convincingly than adoption (Rustemi et al., 2023). Second, the technical work concentrates on higher education diplomas, with vocational certificates treated as an afterthought despite VET's larger share of mid-skill labour markets and its greater exposure to informal and cross-border employment (Chakroun & Keevy, 2018). This imbalance matters because the institutional structure of VET, with many small providers, sectoral bodies, and workplace-based assessment, poses harder issuance and trust problems than a university registry. 2.2 Micro-credentials and the changing unit of certification A second literature examines what is being certified. Micro-credentials, short and focused records of learning, have moved from the margins of online education into national and supranational policy (Bideau & Kearns, 2022; Council of the European Union, 2022). A systematic review of stakeholder perspectives finds that learners want short, career-relevant units, institutions emphasise accreditation and trust, employers demand clarity about the competences a credential certifies, and governments seek employability gains, with these expectations only partially aligned (Varadarajan et al., 2023). In VET specifically, micro-credentials are attractive because they can certify discrete competences acquired through work, recognise prior learning, and be stacked toward fuller qualifications (Gamage & Dehideniya, 2025; Pouliou, 2025). The evidence on employer uptake is mixed. Survey research with human-resource professionals finds recognition of the resume value of micro-credentials alongside persistent doubts about their legitimacy and transferability (Alasmari, 2024). Employer-facing studies of online micro-credentials in soft-skill domains report willingness to treat them as feasible training evidence, conditional on authentic, practice-based assessment (Bruguera et al., 2025). Critical scholarship warns that micro-credentials can function as gig credentials for a gig economy, unbundling qualifications in ways that transfer training costs and risk to individual workers (Wheelahan & Moodie, 2022). The relevance for attestation is direct: as the unit of certification shrinks and multiplies, the volume of documents to be verified grows, and the case for automated, low-cost verification strengthens accordingly (Stanelytė, 2026). Pilot work in East African universities shows that competency-based ePortfolios coupled to digital micro-credentialing can make employability skills visible to employers in contexts where traditional transcripts do not (Maina et al., 2022). 2.3 Credentials as labour-market signals Economic research supplies the theoretical foundation for why attestation matters at all. In signaling terms, a credential has value to the extent that it credibly communicates unobservable productivity. Experimental work with human-resource managers shows that signals of cognitive skills, social skills, and maturity on application documents causally affect interview invitations, with the effective signal differing between apprenticeship applicants and graduates (Piopiunik et al., 2020). Field experiments in the German apprenticeship market demonstrate that employers screen on school-report signals of both cognitive and noncognitive skills, applying thresholds rather than linear rankings (Protsch & Solga, 2015). Most directly relevant to VET attestation, a field experiment in Uganda found that certificates disclosing workers' noncognitive skills improved matching between workers and firms and raised earnings, indicating that credible certification of otherwise unobservable competences has real allocative value in labour markets with weak information infrastructure (Bassi & Nansamba, 2022). The demand side of signaling is itself changing. Analysis of approximately eleven million United Kingdom job postings finds that employers in shortage occupations have begun substituting skill requirements for degree requirements, with measured skills commanding wage premia that rival formal qualifications (Bone et al., 2025). Vacancy-data research shows that machine-readable skill taxonomies now mediate how demand is expressed and matched (Colombo et al., 2019), and reviews of algorithmic recruitment document the spread of automated parsing, shortlisting, and matching alongside concerns about bias, transparency, and explainability (Broecke, 2023). A credential that an algorithm cannot verify and parse is, in such markets, progressively invisible. 2.4 Interoperability, recognition, and the cross-border problem A fourth strand addresses the infrastructures that make credentials meaningful across systems. UNESCO's analysis of digital credentialing argues that digital learning records challenge conventional models of credential evaluation and that their value depends on synergy with qualifications frameworks and quality assurance rather than on technology alone (Chakroun & Keevy, 2018). The European approach illustrates institutionalisation: a common definition and standard data elements for micro-credentials (Council of the European Union, 2022), an infrastructure for digitally sealed credentials linked to Europass, and continuing work on aligning credentials to framework levels (Bideau & Kearns, 2022). Recent scholarship extends this agenda to artificial intelligence, examining whether large language models can assist the comparison and levelling of qualifications across frameworks while cautioning that human oversight and accountable governance remain necessary (Keevy et al., 2026). Learner-centred architectures form part of the same conversation. Self-sovereign identity systems issue verifiable credentials to wallets under the worker's control, allowing selective disclosure to employers without repeated recourse to the issuer (Grech et al., 2021; Chan et al., 2025). Surveys of these systems identify unresolved questions of governance, key management, revocation, and issuer credibility, and note that the gap between promise and practice remains wide in education (Grech et al., 2021). Finally, an equity literature qualifies the entire agenda. Studies of digital technology access in vocational schooling show that material, motivational, and skills-related access conditions vary sharply by geography (Habibi et al., 2023), and analyses of the digital divide in developing economies and the G20 argue that skills and infrastructure gaps reproduce exclusion unless deliberately countered (Chetty et al., 2018; Niwamanya et al., 2025). Digital attestation inherits these constraints: a verification system usable only by connected, digitally literate actors can disadvantage precisely the workers and small employers that vocational systems serve. The equity qualification extends inside wealthy systems as well. Access research in Indonesian vocational schools, based on responses from over thirteen hundred participants, found that motivational access predicted skills access more strongly than material access predicted it, and that geography differentiated every access variable measured (Habibi et al., 2023). The finding matters for attestation design because it locates the constraint not only in devices and connectivity but in the confidence and capability of teachers and learners, which infrastructure spending alone does not repair (Chetty et al., 2018). Read together, the four strands show mature but siloed knowledge. What is missing is an integrated account that connects verification technology to certification units, signaling economics, and recognition infrastructure, and that is specific to the vocational sector. The present article constructs that account. 3. Method 3.1 Design and rationale The study is an integrative review with a conceptual contribution. An integrative design was chosen over a systematic review because the relevant evidence is distributed across heterogeneous fields, computer science, education policy, and labour economics, whose research questions, methods, and quality conventions differ too much for a single effect-oriented synthesis, and because the aim is theory construction rather than exhaustive enumeration. A purely conceptual paper without a structured evidence base was rejected because the framework's claims are intended to be anchored in verifiable findings. The research questions stated in the Introduction guide the design: the first requires organised coverage of the evidence in each dimension; the second requires an explicit procedure for deriving an integrating structure from that evidence. 3.2 Search and selection Literature was identified between June and August 2026 through structured queries of scholarly databases (Consensus, drawing on Semantic Scholar, Scopus, and related indexes) complemented by targeted web searches of institutional repositories (UNESCO, OECD, EUR-Lex). Eight query families covered the constructs in the title: blockchain and certificate verification; micro-credentials and vocational recognition; digital credentials and cross-border mobility; self-sovereign identity in education; labour-market signaling and skill certification; skills-based hiring; algorithmic job matching; and digital-divide constraints on vocational technology adoption. Inclusion criteria were: peer-reviewed journal articles, or authoritative institutional and legal documents, published principally between 2015 and 2026; direct relevance to at least one framework dimension; and confirmable bibliographic identity. Seminal earlier work was admitted where the argument required it. Exclusion criteria were: purely promotional literature, papers whose metadata could not be verified against the original record, and technical papers with no implication beyond implementation detail. Every retained source was verified against its digital object identifier or publisher record before citation; twenty-nine sources satisfied all criteria and form the evidence base. 3.3 Analytical procedure Analysis proceeded in three steps. First, retained sources were coded inductively by the principal problem each addresses: authenticity, meaning, recognition, or use. Second, the coded groups were examined for the dependencies among them, asking what each group presupposes from the others; this dependency analysis generated the layered structure presented in Section 4, in which each layer supplies a condition the next layer assumes. Third, the draft framework was tested against the evidence by checking, for each layer, whether the reviewed studies support the claimed function and whether documented failures of attestation initiatives correspond to missing layers. Claims in Section 4 are attributed to sources at the strength the underlying study warrants, and framework-level statements are marked as conceptual. 3.4 Rigor and boundaries Several measures limited bias. Searches deliberately included critical scholarship on micro-credentials and algorithmic hiring so that the framework would not encode a promotional reading of the technology. No source was cited from memory; all bibliographic records were confirmed against the original publication. The review's boundaries are acknowledged: it synthesises published research and policy documents rather than new fieldwork; the policy material overrepresents the European infrastructure because that is where institutionalisation is most documented; and the framework's validity is conceptual until tested empirically. These boundaries are revisited in the Limitations subsection. 4. A Four-Layer Framework for Digital Attestation of Vocational Certificates The dependency analysis yields a framework of four layers, each supplying a condition that the next assumes, with equity operating as a cross-cutting condition on all four. Figure 1 summarises the framework, and the subsections develop each layer with its supporting evidence. Table 1 synthesises the principal attestation mechanisms documented in the literature and locates each within the framework. 4.1 Layer 1: Cryptographic trust The foundational layer secures the authenticity and integrity of the certificate itself. The reviewed systems achieve this through digital signatures, cryptographic hashing, and anchoring on distributed ledgers, whether permissioned architectures operated by education authorities (Khan et al., 2021; Rani et al., 2024), public chains with decentralised storage (Said et al., 2025), or standards-based deployments such as Blockcerts (Capece et al., 2020). The function of this layer is precise: it converts verification from a correspondence process into a computation, making forgery detectable and checking effectively free at the margin. Evidence indicates the layer is technically mature; performance evaluations report workable throughput and latency (Rani et al., 2024), and comparative analyses document advantages over centralised registries in tamper-resistance and availability (Said et al., 2025). Two design tensions within the layer deserve emphasis because they shape everything above it. The first concerns governance. Permissioned architectures concentrate control in a consortium of issuers, which suits national authorities but reproduces at the technical level the institutional boundaries that cross-border attestation is meant to dissolve (Khan et al., 2021). Public-chain designs distribute trust more widely but leave open who vouches for the issuer whose signature the chain merely preserves (Said et al., 2025). The second concerns permanence and privacy. Immutability is the property that makes ledger-anchored certificates trustworthy, yet credentials sometimes must be revoked, corrected, or removed, and personal data written into permanent structures raises data-protection questions that implementations resolve by keeping credential content off-chain and anchoring only hashes (Rustemi et al., 2023; Chan et al., 2025). Neither tension is fatal, but both show that even the most technical layer embeds institutional choices. The layer's characteristic failure mode is isolation. A cryptographically perfect certificate whose content no external system can interpret merely proves that an unintelligible document is authentic. This is the condition the second layer addresses. 4.2 Layer 2: Semantic interoperability The second layer makes the content of the certificate machine-readable and comparable. Its instruments are data standards for credentials, skills taxonomies that decompose qualifications into competences (Colombo et al., 2019), and the standard elements defined for micro-credentials in the European approach (Council of the European Union, 2022). The layer matters because labour-market intermediaries increasingly operate on structured skill data rather than document images (Broecke, 2023), and because the shrinking unit of certification multiplies the number of records that must be interpreted without human mediation (Stanelytė, 2026). For VET, semantic work is harder than for degrees: vocational competences are more granular, more contextual, and less standardised across countries (Chakroun & Keevy, 2018). Emerging research proposes artificial intelligence as an aid to comparing and levelling qualifications across frameworks, with explicit warnings that such systems require human oversight and accountable governance (Keevy et al., 2026). The layer also governs stackability. If micro-credentials are to accumulate toward fuller qualifications, as both policy and stakeholder research envisage (Council of the European Union, 2022; Varadarajan et al., 2023), the units must share data elements that allow a receiving institution or employer to compute what a given combination amounts to. Without common semantics, stacking is a metaphor rather than an operation. Conceptually, this layer converts an authentic certificate into usable information; without it, cryptographic trust terminates at the document boundary. 4.3 Layer 3: Institutional recognition The third layer connects verified, interpretable credentials to the institutions that confer standing: quality assurance bodies, qualifications frameworks, professional regulators, and recognition conventions. The literature is consistent that this layer, not technology, is the binding constraint on cross-border value. UNESCO's analysis argues that digital credentials acquire portability only in synergy with qualifications frameworks and quality assurance (Chakroun & Keevy, 2018), and the European instruments operationalise this by tying digitally sealed credentials to framework levels and common data elements (Bideau & Kearns, 2022; Council of the European Union, 2022). Stakeholder research on micro-credentials shows that institutional accreditation is precisely what employers and institutions demand before trusting new credential forms (Varadarajan et al., 2023; Alasmari, 2024). Self-sovereign architectures redistribute agency within this layer rather than abolishing it: worker-held credentials still derive authority from recognised issuers, and surveys identify issuer credibility and governance as the unresolved core of such systems (Grech et al., 2021; Chan et al., 2025). The characteristic failure mode of initiatives that neglect this layer is the pilot that verifies flawlessly but confers no standing an employer or regulator is obliged to honour. 4.4 Layer 4: Labour-market utilisation The final layer is where attested certificates do economic work. Signaling research establishes the mechanism: credible certification of otherwise unobservable competences improves matching and can raise earnings, with the strongest experimental evidence coming from certification of noncognitive skills in a developing labour market (Bassi & Nansamba, 2022), and complementary evidence showing employers act on documented skill signals at labour-market entry (Protsch & Solga, 2015; Piopiunik et al., 2020). The demand side is shifting toward exactly the granular, skill-level information that digitally attested credentials can carry: employers in shortage fields increasingly specify skills rather than degrees (Bone et al., 2025), and algorithmic intermediaries parse, rank, and match on structured credential data (Broecke, 2023). Micro-credential pilots that make employability skills visible to employers demonstrate the layer in operation (Maina et al., 2022), while employer scepticism about legitimacy (Alasmari, 2024) shows what happens when the upper layer runs ahead of institutional recognition. The critical scholarship belongs in this layer too: if unbundled credentials feed platforms that intensify contingent work, attestation may amplify precarity rather than mobility (Wheelahan & Moodie, 2022). The framework does not resolve that normative question, but it locates it where it arises. 4.5 The cross-cutting equity condition Every layer presupposes access: to devices, connectivity, digital skills, and institutional presence. Evidence from vocational schooling shows access conditions vary sharply within single countries (Habibi et al., 2023), and analyses of developing economies document infrastructure, cost, and educator-capability constraints on vocational digitalisation (Niwamanya et al., 2025), with digital-skills strategy identified as the remedial lever at policy level (Chetty et al., 2018). For a framework aimed at borderless workforces the implication is sharp: the workers with most to gain from portable attested credentials, migrants and workers in low-income systems, are those most exposed to exclusion from the infrastructure that carries them. Equity is therefore modelled as a condition on all four layers rather than a fifth layer, since failure of access disables the stack wherever it occurs. Table 1 Digital Attestation Mechanisms in the Reviewed Literature, Located Within the Framework Mechanism Primary function Trust anchor Framework layer(s) Key evidence Principal limitation Permissioned-ledger attestation by education authorities Tamper-evident issuance and verification of certificates Consortium of authorised institutions 1 Khan et al. (2021); Rani et al. (2024) Governance concentrated in issuing consortium; limited cross-border reach Public-chain certificate systems with decentralised storage Country-scale verification across education levels Public blockchain 1 Said et al. (2025) Adoption and legal standing untested beyond pilot context Standards-based digital diplomas (e.g., Blockcerts) Institution-issued verifiable diplomas Open standard plus public chain 1–2 Capece et al. (2020) Single-institution pilots; weak link to qualifications frameworks Self-sovereign identity with verifiable credentials Worker-held, selectively disclosable credentials Decentralised identifiers; recognised issuers 1–3 Grech et al. (2021); Chan et al. (2025) Issuer credibility, revocation, and governance unresolved Micro-credentials with standard data elements Granular, stackable certification of competences Accredited issuers; common definitions 2–3 Council of the European Union (2022); Varadarajan et al. (2023) Employer trust and transferability still contested (Alasmari, 2024) Framework-linked digital credential infrastructure (Europass model) Cross-border legibility and recognition Qualifications frameworks; sealed credentials 2–3 Bideau & Kearns (2022); Chakroun & Keevy (2018) Coverage concentrated in Europe; VET granularity challenges Skill-certificate signaling into hiring platforms Matching workers to jobs on verified skills Employer-recognised certification 4 Bassi & Nansamba (2022); Bone et al. (2025) Depends on all lower layers; algorithmic bias risks (Broecke, 2023) Note. Layers: 1 = cryptographic trust; 2 = semantic interoperability; 3 = institutional recognition; 4 = labour-market utilisation. The table synthesises mechanisms as characterised in the cited sources; classifications by layer are the author's analytical judgement. Figure 1. The four-layer framework of digital attestation for vocational education certificates. Each layer supplies a condition the next layer assumes; equity in access operates as a cross-cutting condition on all layers. Source: author's elaboration based on the reviewed literature. 5. Discussion 5.1 Theoretical implications The framework's central claim is that digital attestation is a stack of dependent conditions, not a technology. This reframing explains a pattern the separate literatures record but do not account for: verification systems that work and are not used. In the reviewed corpus, technically successful systems stall at the semantic layer, when their certificates cannot be interpreted by external systems, or at the institutional layer, when verification confers no recognised standing (Rustemi et al., 2023; Grech et al., 2021). Conversely, signaling research presupposes credibility that only the lower layers can supply at scale; the experimental gains from skill certification observed by Bassi and Nansamba (2022) depended on a certification the employers in that market treated as credible. Connecting these literatures places credential verification inside the economics of information rather than alongside it: the lower layers of the stack are, in signaling terms, technologies for reducing the cost of credibility. A second theoretical implication concerns the direction of dependence. The layers are ordered by presupposition, not by importance, and value flows downward as well as up. Employer demand for granular skill information (Bone et al., 2025) creates the economic case that justifies investment in semantic standards; recognition instruments give issuers a reason to adopt cryptographic infrastructure they would not adopt for its own sake (Bideau & Kearns, 2022). The framework therefore predicts that attestation ecosystems consolidate fastest where a strong fourth-layer pull, acute skill shortages or high verification costs, meets an existing third-layer architecture, and slowest where technology is supplied without either. The uneven geography of the reviewed initiatives is consistent with this prediction, though it does not test it. The framework also clarifies what is genuinely different about the vocational case. VET certificates certify granular, contextual competences issued by heterogeneous providers, which raises the difficulty of the semantic and institutional layers relative to university degrees (Chakroun & Keevy, 2018). The literature's concentration on higher education has therefore solved the easier version of the problem. Treating VET as the primary case, as this article does, exposes the semantic layer as the sector's decisive bottleneck. 5.2 Practical and policy implications For policymakers and credentialing bodies the framework functions as a sequencing device. Investment in cryptographic infrastructure without parallel work on data standards and framework alignment produces authentic but illegible certificates; the reviewed European instruments indicate the alternative order, in which common definitions and framework linkage precede or accompany technical deployment (Council of the European Union, 2022; Bideau & Kearns, 2022). For VET providers, the immediate implication is that issuing digitally sealed credentials aligned to standard data elements does more for graduates' mobility than joining any particular ledger. For employers and platform operators, verified credential data offers a defensible input to skills-based hiring at a moment when degree filters are weakening (Bone et al., 2025), subject to the governance concerns algorithmic recruitment already raises (Broecke, 2023). For development policy, the equity condition implies that attestation programmes in low-income systems must budget for access and capability, not only for technology, and that partnerships with enterprises can carry part of that burden (Niwamanya et al., 2025; Chetty et al., 2018). The framework also suggests a division of labour among actors that current initiatives blur. Ledger infrastructure and wallet standards are public-good candidates best provided once and shared, as the cost structure of the reviewed systems implies (Rani et al., 2024). Semantic assets, taxonomies, data schemas, and framework mappings require the participation of sectoral bodies that understand vocational competence, and they decay without maintenance as occupations change (Colombo et al., 2019). Recognition, finally, cannot be outsourced to technology at all: quality assurance and framework alignment remain the province of the institutions the third layer describes, and initiatives that promise to bypass them substitute verification for trust rather than producing it (Chakroun & Keevy, 2018; Grech et al., 2021). Programmes can be audited against this division: an attestation project unable to state which actor sustains each layer is incomplete in a way the framework makes visible in advance. 5.3 Limitations Three limitations bound these claims. First, the framework is derived from published research and policy documents, not from new empirical observation; it organises existing evidence and its validity is correspondingly conceptual. Second, the evidence base is uneven: the technical layer is documented largely through prototypes and pilots rather than longitudinal adoption studies, and the policy material overrepresents the European infrastructure, so the framework's institutional layer may transfer imperfectly to systems with weaker qualifications architectures. Third, the labour-market layer rests on signaling studies conducted with instruments other than blockchain-attested certificates; the inference that attested digital credentials will reproduce those signaling gains is theoretically grounded but empirically untested. These limitations follow directly from the method and define the research agenda below. 5.4 Future research The priority is empirical work at the junctions between layers. Field experiments issuing verifiably attested VET credentials into real hiring processes would test whether cryptographic verification changes employer behaviour, extending the design of Bassi and Nansamba (2022) to digital instruments. Comparative institutional research should examine how attestation initiatives fare in qualification systems outside Europe, where the recognition layer is thinner. Design research on semantic interoperability for vocational competences, including the accountable use of large language models in qualification comparison (Keevy et al., 2026), addresses the bottleneck this review identifies. Finally, longitudinal study of workers using self-sovereign wallets across borders would provide the first direct evidence on whether learner-held attestation supports, or merely accompanies, labour mobility. 6. Conclusion This article set out to explain what digital attestation must accomplish for vocational education certificates to function in algorithmically mediated, cross-border labour markets. Integrating evidence from computer science, education policy, and labour economics, it developed a four-layer framework in which cryptographic trust, semantic interoperability, institutional recognition, and labour-market utilisation each supply conditions the next layer assumes, with equitable access conditioning the whole. The review found the technical layer mature, the semantic and institutional layers decisive and comparatively neglected for VET, and the labour-market layer theoretically well founded but empirically untested for attested digital instruments. The framework converts a fragmented literature into a sequencing logic for policy and a set of testable junctions for research. As certification fragments into smaller units and hiring migrates onto data-driven platforms, the systems that attest vocational competence will increasingly determine whose skills are visible, and therefore whose work is possible, across borders. Declarations Funding. This research received no external funding. Conflicts of Interest. The author declares no conflict of interest. Ethics Statement. This study is an integrative review of published literature and involved no human participants, personal data, or animal subjects; ethical approval was therefore not required. Data Availability. No new data were created or analysed in this study. All sources reviewed are publicly available through the references listed. References Alasmari, T. (2024). Reshaping vocational training: A study on the recognition of micro-credentials in job markets. Education + Training, 66(2/3), 233–251. https://doi.org/10.1108/ET-07-2023-0282 Bassi, V., & Nansamba, A. (2022). Screening and signalling non-cognitive skills: Experimental evidence from Uganda. The Economic Journal, 132(642), 471–511. https://doi.org/10.1093/ej/ueab071 Bideau, Y.-M., & Kearns, T. (2022). A European approach to micro-credentials for lifelong learning and employability. Journal of European CME, 11(1), Article 2147288. https://doi.org/10.1080/21614083.2022.2147288 Bone, M., González Ehlinger, E., & Stephany, F. (2025). Skills or degree? The rise of skill-based hiring for AI and green jobs. Technological Forecasting and Social Change, 214, Article 124042. https://doi.org/10.1016/j.techfore.2025.124042 Broecke, S. (2023). Artificial intelligence and labour market matching (OECD Social, Employment and Migration Working Papers No. 284). OECD Publishing. https://doi.org/10.1787/2b440821-en Bruguera, C., Pagés, C., Peters, M., & Fitó, À. (2025). Micro-credentials and soft skills in online education: The employers' perspective. Distance Education, 46(1), 56–76. https://doi.org/10.1080/01587919.2024.2435645 Capece, G., Levialdi Ghiron, N., & Pasquale, F. (2020). Blockchain technology: Redefining trust for digital certificates. Sustainability, 12(21), Article 8952. https://doi.org/10.3390/su12218952 Castro, R. Q., & Au-Yong-Oliveira, M. (2021). Blockchain and higher education diplomas. European Journal of Investigation in Health, Psychology and Education, 11(1), 154–167. https://doi.org/10.3390/ejihpe11010013 Chakroun, B., & Keevy, J. (2018). Digital credentialing: Implications for the recognition of learning across borders. UNESCO. https://doi.org/10.54675/SABO8911 Chan, W., Gai, K., Yu, J., & Zhu, L. (2025). Blockchain-assisted self-sovereign identities on education: A survey. Blockchains, 3(1), Article 3. https://doi.org/10.3390/blockchains3010003 Chetty, K., Aneja, U., Mishra, V., Gcora, N., & Josie, J. (2018). Bridging the digital divide in the G20: Skills for the new age. Economics: The Open-Access, Open-Assessment E-Journal, 12(2018-24), 1–20. https://doi.org/10.5018/economics-ejournal.ja.2018-24 Colombo, E., Mercorio, F., & Mezzanzanica, M. (2019). AI meets labor market: Exploring the link between automation and skills. Information Economics and Policy, 47, 27–37. https://doi.org/10.1016/j.infoecopol.2019.05.003 Council of the European Union. (2022). Council Recommendation of 16 June 2022 on a European approach to micro-credentials for lifelong learning and employability (2022/C 243/02). Official Journal of the European Union, C 243, 10–25. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32022H0627(02) Gamage, K. A. A., & Dehideniya, S. C. P. (2025). Unlocking career potential: How micro-credentials are revolutionising higher education and lifelong learning. Education Sciences, 15(5), Article 525. https://doi.org/10.3390/educsci15050525 Grech, A., Sood, I., & Ariño, L. (2021). Blockchain, self-sovereign identity and digital credentials: Promise versus praxis in education. Frontiers in Blockchain, 4, Article 616779. https://doi.org/10.3389/fbloc.2021.616779 Habibi, A., Sofyan, S., & Mukminin, A. (2023). Factors affecting digital technology access in vocational education. Scientific Reports, 13, Article 5682. https://doi.org/10.1038/s41598-023-32755-6 Keevy, J., Bateman, A., Manoharan, H. E., Molokwane, P., Lastra, R., Mlotshwa, H., Paterson, A., Ramchurun, R., Ravaioli, S., Shiohira, K., von Doetinchem, S., & Warren, K. (2026). Beyond qualifications frameworks: Large language models and the future of global skills recognition. European Journal of Open, Distance and E-Learning, 28(1), Article 3. https://doi.org/10.65043/eurodl.162 Khan, A. A., Laghari, A. A., Shaikh, A. A., Bourouis, S., Mamlouk, A. M., & Alshazly, H. (2021). Educational blockchain: A secure degree attestation and verification traceability architecture for Higher Education Commission. Applied Sciences, 11(22), Article 10917. https://doi.org/10.3390/app112210917 Maina, M. F., Guàrdia Ortiz, L., Mancini, F., & Martinez Melo, M. (2022). A micro-credentialing methodology for improved recognition of HE employability skills. International Journal of Educational Technology in Higher Education, 19, Article 10. https://doi.org/10.1186/s41239-021-00315-5 Niwamanya, G., Zhong, Z., Matola, C. A., Eliachu, D., Ravindra, S., & Fahn, J. B. (2025). Bridging the digital divide through enterprise-TVET provider partnerships: Strategic insights on digital transformation in TVET in developing nations. Vocation, Technology & Education, 2(2), Article e11. https://doi.org/10.54844/vte.2025.0977 Piopiunik, M., Schwerdt, G., Simon, L., & Woessmann, L. (2020). Skills, signals, and employability: An experimental investigation. European Economic Review, 123, Article 103374. https://doi.org/10.1016/j.euroecorev.2020.103374 Pouliou, A. (2025). Deepening the focus on microcredentials—labour market sector relevance, quality and inclusion: A critical commentary. Distance Education, 46(1), 113–122. https://doi.org/10.1080/01587919.2025.2453457 Protsch, P., & Solga, H. (2015). How employers use signals of cognitive and noncognitive skills at labour market entry: Insights from field experiments. European Sociological Review, 31(5), 521–532. https://doi.org/10.1093/esr/jcv056 Rani, P., Sachan, R. K., & Kukreja, S. (2024). Educert-chain: A secure and notarized educational certificate authentication and verification system using permissioned blockchain. Cluster Computing, 27(7), 10169–10196. https://doi.org/10.1007/s10586-024-04469-5 Rustemi, A., Dalipi, F., Atanasovski, V., & Risteski, A. (2023). A systematic literature review on blockchain-based systems for academic certificate verification. IEEE Access, 11, 64679–64696. https://doi.org/10.1109/ACCESS.2023.3289598 Said, S. H., Sinde, R. S., Kosia, E. M., & Dida, M. A. (2025). A comprehensive blockchain-based system for educational qualifications management and verification to counter forgery. IEEE Access, 13, 31562–31589. https://doi.org/10.1109/ACCESS.2025.3542545 Stanelytė, D. (2026). Recognising employees' competences through verifiable micro-credentials: Aligning higher education and employers' needs in the digital era. Entrepreneurship and Sustainability Issues, 13(3), 143–158. https://doi.org/10.9770/n3433346968 Varadarajan, S., Koh, J. H. L., & Daniel, B. K. (2023). A systematic review of the opportunities and challenges of micro-credentials for multiple stakeholders: Learners, employers, higher education institutions and government. International Journal of Educational Technology in Higher Education, 20(1), Article 13. https://doi.org/10.1186/s41239-023-00381-x Wheelahan, L., & Moodie, G. (2022). Gig qualifications for the gig economy: Micro-credentials and the 'hungry mile'. Higher Education, 83(6), 1279–1295. https://doi.org/10.1007/s10734-021-00742-3 Hashtags: #SDG4 #QualityEducation #SDG8 #DecentWork #SDG10 #ReducedInequalities #DigitalCredentials #VocationalEducation #BlockchainAttestation #MicroCredentials #LabourMobility #SkillsRecognition #VerifiableCredentials #SmartJobMarkets #U7YJournal