The study uses an integrative review to assemble and synthesize heterogeneous scholarly and sector evidence, linking framework choices and governance patterns to context constraints on enterprise architecture (EA) in higher education. The integrative approach is well-suited to combining conceptual, methodological, and empirical contributions to derive design principles, governance implications, and a context-adapted pathway for Yemen. ^{21, 22} It allows inclusion of diverse evidence types, peer-reviewed journal articles, conference proceedings, and high-credibility sector outputs, capturing both scholarly and applied practice perspectives.

To enhance transparency of the selection process, the studies’ identification diagram flow diagram summarizes identification, screening, eligibility, and inclusion ( Figure 1). Core characteristics of the 30 included studies are summarized in Table 1, with full extraction data provided in the supplementary literature review matrix. ²³

We defined the scope and boundaries a priori to concentrate on EA in higher education, specifically addressing its implications for digital transformation, governance, interoperability, and fragile context transferability. We prioritized studies that: •

Applied or adapted an EA framework to higher education;

Quality and credibility were assessed using transparent criteria: clarity of EA framework/method, documentation of governance arrangements, methodological transparency (design, data sources), and validation (pilots, surveys, expert review).

•

RQ1: How are EA frameworks and methods selected, adapted, and implemented in higher education institutions and multi-institutional settings?

Together, these questions address the overarching aim of synthesizing evidence on EA in higher education to inform governance and implementation strategies suited to fragile environments.

An iterative, multi-source, snowballing search was conducted across major academic databases and publisher platforms (Scopus, IEEE Xplore, ScienceDirect, SpringerLink, and Google Scholar), supplemented by targeted searches of sector proceedings and community repositories (e.g., EUNIS/EPiC, ACM/IEEE) to identify scholarly and practice-relevant evidence.

Search strategies enforced two mandatory concept clusters (“enterprise architecture”) AND (“higher education” OR universit*), plus a contextual cluster capturing digital transformation/IT governance, multi-institutional/sectoral, or federated settings. Filters applied: English, 2010–2025. Optional qualifiers (“fragile”, “low-resource”, “conflict-affected”, “Yemen”) were tested in some iterations to explore contextual coverage without unduly narrowing recall.

Advanced search fields (Title/Abstract/Keywords) were used where possible, with truncation and phrase matching applied (e.g., “enterprise architecture”, universit*, sector*, interoperab*). This combination of structured database searching and targeted venue scanning balanced coverage breadth with topical specificity.

2.4.1 Inclusion criteria: •

Domain: Higher-education institutions (institutional, multi-institutional, or sector-level) or sector frameworks explicitly applicable to HE;

2.4.2 Exclusion criteria: •

Non-HE contexts without explicit transferability;

Screening followed a two-stage process: •

Title/Abstract screening for core relevance;

The flow of records through these stages is illustrated in Figure 1 (Studies identification diagram). A total of ( n=220) records were identified through applying the searching strategy on the identified information sources. After ( n=25) duplicates and ( n=10) non-HE-relevant records were removed, ( n=195) records were screened for relevance requirements, of which ( n=158) were excluded. The remaining ( n=37) full-text articles were assessed for eligibility. Of these, ( n=7) reports were excluded for reasons such as insufficient details of EA implementation, focusing on a single-department scope rather than the institutional or multi-agency scope, and lacking multi-institutional or strategic components with limited transferability to fragile and multi-institutional contexts. These exclusions resulted in a final corpus of ( n=30) studies that met the inclusion criteria and were included in the integrative review.

Records were exported in BibTeX format, merged, and deduplicated before screening. For works with multiple manifestations (e.g., publisher and repository copies), a canonical version was retained. Each included study was assigned a unique ID in the literature review matrix. ²³ Table 1 summarizes data extraction dimensions within the analytical framework for the included studies.

Data extraction captured: •

Study characteristics (country/region, year, type, scope);

Descriptive counts (e.g., TOGAF prevalence) supported pattern detection; narrative synthesis integrated qualitative insights and case comparisons to address the three RQs.

The data synthesis followed a thematic analysis approach. An initial deductive codebook was created based on the review’s research questions (e.g., EA frameworks, governance models, implementation strategies). This was supplemented by an inductive, open coding process to capture emergent themes and unexpected findings from the included studies. Both authors independently coded a sample of articles ( n=5) and discussed the results to refine the codebook and ensure consistent application. The final coded data, managed in the literature review matrix, ²³ was then synthesized narratively to identify patterns, contrasts, and relationships across the corpus.

Applying integrative review guidance, ²¹ we conducted a credibility appraisal to evaluate the included studies. A pragmatic three-tier rating was used to inform our synthesis rather than for strict exclusion. The appraisal was conducted by both authors utilizing a calibrated consensus-based methodology to ensure reliability. Both authors independently utilized a predetermined rubric to assess source credibility (peer-review status and venue reputation), methodological transparency (clarity of design and evidence), and transferability (applicability to other contexts) on a subset of the studies ( n=10, 33%) to establish a common understanding of the criteria. Subsequent to the calibration phase, they jointly examined all 30 studies, addressed any minor rating discrepancies, and achieved a final consensus on the classification of each study. This iterative process guaranteed the consistent and reliable application of the appraisal criteria throughout the entire corpus. ²⁵ This process resulted in the classification of studies into three distinct categories: •

High Credibility: Studies demonstrating robust design and evidence, typically from highly ranked, peer-reviewed venues;

Following this appraisal, our final corpus consisted of ( n=14) high credibility studies, ( n=9) medium credibility studies, and ( n=7) high relevance/non-traditional sources. This approach allowed for a comprehensive synthesis that balanced methodological rigor with the practical and conceptual insights essential for an integrative review. A full rating with justification is provided in the supplementary literature review matrix. ²³

The contextual constraints reported in the studies (e.g., autonomy/siloing, legacy fragmentation, resource gaps, and regulation) were mapped to Yemen’s sector realities. Convergent mechanisms across studies, such as federated governance paired with reference architectures, pilot-led phasing, and capability-driven planning, informed the pathway for Yemen presented in Section 4.

This integrative review was conducted with a structured and transparent approach consistent with good-practice standards for evidence synthesis. ^{21, 22} Study handling is documented through an identification-to-inclusion diagram ²⁴ summarizes the identification, screening, eligibility, and inclusion stages ( Figure 1). Detailed data extraction and coding were performed exclusively for the 30 studies meeting the final inclusion criteria, as summarized in Table 1 and in the full supplementary matrix. ²³ Records excluded in earlier screening stages were not retained beyond aggregate counts and summary exclusion reasons, in line with the review’s integrative design and our transparency commitments. The final sample of 30 studies represents a purposive selection of relevant EA implementations in higher education, rather than an exhaustive enumeration of all such initiatives globally.

Potential limitations include: •

Publication bias: peer-reviewed literature may over-represent successful EA implementations relative to unsuccessful or discontinued initiatives;

Steps taken to mitigate these risks include broad, multi-database searching supplemented with targeted sector proceedings; inclusion of diverse evidence types; and standardized extraction and qualitative appraisal of study characteristics, evidence strength, and transferability.

Note on language refinement: To assist with clarity of English expression, the artificial intelligence tools of quillbot (online tool at quillbot.com) were used selectively for phrasing improvements. The authors critically reviewed, edited, and approved all AI-refined text to ensure accuracy, preserve intended meaning, and maintain research integrity. The tool did not generate any new research ideas or original data.

The final dataset encompasses 30 studies from 20 distinct national or multi-country contexts, demonstrating substantial global interest in EA adoption within higher education. The geographic distribution of this research, summarized by region in Figure 2, reveals a significant concentration of studies originating from Southeast Asia (9 studies). The above pattern is followed by notable research activity in Europe (5 studies) and the Middle East (5 studies), with the remaining work spanning a diverse range of other countries and multi-country contexts (11 studies). At a national level, the concentration in Southeast Asia is primarily driven by extensive research from Indonesia (6 studies), with other key country-specific contributions including those from Finland, Saudi Arabia, Egypt, and Thailand (2 studies each), alongside individual papers from contexts such as Norway, Poland, Malaysia, and Libya.

Study methodologies varied considerably, with empirical case studies comprising the largest category (17 studies, 56.7%), followed by conceptual frameworks (8 studies, 26.7%), systematic reviews (3 studies, 10%), and national program evaluations (2 studies, 6.7%). The temporal distribution shows marked acceleration in research activity after 2019, with 18 studies (60%) published in the post-pandemic period, suggesting heightened recognition of EA’s strategic importance for institutional resilience and digital transformation. ³

Credibility assessment revealed that nearly half (46.7%) of our sources are of high methodological quality, providing a robust, peer-reviewed core to the findings. A combined 70% (high and medium credibility) of our sources are traditionally peer-reviewed, demonstrating that this review is firmly grounded in established academic literature. The 23.3% categorized as “high relevance/non-traditional source”. This group contains some of the most critical papers for the topic included for their strategic value, such as the Libyan, Colombian, and Egyptian models. ^{5, 11, 26} This distribution ensures robust empirical foundations for our analysis while acknowledging varying levels of methodological sophistication across the corpus.

3.2.1 Framework selection dominance

TOGAF emerges as the overwhelmingly dominant framework, utilized in 26 out of 30 studies (86.7%), confirming its position as the de facto standard for higher education EA initiatives globally. ⁹ This dominance reflects TOGAF’s comprehensive methodology, vendor neutrality, and proven adaptability to complex organizational environments characteristic of higher education institutions. The distribution of framework adoption across the reviewed studies demonstrates clear patterns that inform strategic decisions for fragile state contexts.

This distribution confirms TOGAF’s versatility in complex academic environments while highlighting the emergence of context-specific adaptations in challenging operational conditions. As shown in Table 2, this dominance is particularly pronounced in multi-institutional contexts and proven adaptability to complex academic environments, while hybrid approaches in 23.3% of studies suggest contextual adaptation requirements.

The Architecture Development Method (ADM) within TOGAF proved particularly valuable for higher education contexts, with 18 studies specifically referencing its phased approach for managing transformation complexity. ²⁷ Studies consistently highlighted TOGAF’s strength in providing structured guidance for baseline assessment, target architecture design, and transition planning critical capabilities for institutions managing legacy system modernization. ⁸

To provide a clearer comparative perspective, Table 2b synthesizes the dominant implementation patterns observed across different geographical regions.

3.2.2 Alternative and hybrid approaches

While TOGAF dominates, alternative frameworks show contextual relevance. The Zachman Framework appeared in four studies (13.3%), primarily valued for its multi-stakeholder perspective and comprehensive modeling capabilities. ^{28, 29} Enterprise Architecture Planning (EAP) methodology gained prominence in 3 studies focusing on business-driven transformation, particularly in resource-constrained environments. ³⁰

Seven studies (23.3%) employed hybrid approaches, combining elements from multiple frameworks to address specific institutional or national requirements. Finland’s Digivisio 2030 program exemplified this trend, integrating TOGAF, ArchiMate, and Business Process Model and Notation (BPMN) within a custom interoperability model designed for multi-institutional collaboration. ¹⁰ Similarly, Egypt’s SECC-SURA framework combined TOGAF with Control Objectives for Information and Related Technology 5 (COBIT5) and IT for IT (IT4IT) to create a comprehensive smart university reference architecture. ¹²

Implementation scope analysis reveals significant variation in ambition and coordination levels. Institutional-level implementations comprised 18 studies (60%), focusing on single university transformation initiatives. Multi-institutional approaches appeared in 8 studies (26.7%), while national or sectoral implementations represented only 4 studies (13.3%).

Federated governance models emerged as the preferred approach for multi-institutional contexts, combining central coordination with institutional autonomy. Finland’s Digivisio 2030 program demonstrated this model’s effectiveness, achieving nationwide interoperability while preserving university independence through shared standards and voluntary participation mechanisms. ¹⁰ Colombia’s Colombian Higher Education Enterprise Architecture (CHE2A) framework similarly balanced national policy alignment with institutional flexibility through reference architectures and collaborative governance structures. ¹¹

3.4.1 Quantified benefits

Studies reporting quantifiable outcomes demonstrated substantial improvements in operational efficiency and stakeholder satisfaction. Indonesia’s UiTM implementation achieved a 25% improvement in student satisfaction, a 30% reduction in administrative processing time, and a 20% increase in faculty satisfaction. ⁴

Cost optimization emerged as a consistent benefit, with institutions reporting 20-40% reductions in IT operational expenses through elimination of redundant systems and improved resource utilization. ^{2, 31}

EA implementations consistently enabled expanded educational access through improved digital infrastructure. Finland’s Digivisio 2030 created unified authentication and learner services supporting seamless cross-institutional mobility. ¹⁰ Similar access improvements appeared across contexts, with average gains of 15-35% in student service accessibility and 20-30% improvements in administrative efficiency.

3.4.2 Strategic Capabilities Enhancement

EA implementations consistently delivered improved system integration, with 28 studies (93.3%) reporting enhanced data sharing and process coordination capabilities. ^{14, 32} This benefit proved especially valuable for multi-institutional environments requiring coordinated service delivery, directly supporting national education system strengthening objectives central to ICT4D frameworks.

Enhanced analytical capabilities and data-driven governance appeared in 25 studies (83.3%), with institutions reporting improved strategic planning and resource allocation decisions. ^{30, 33} The integration of business intelligence capabilities with EA frameworks particularly strengthened institutional capacity for evidence-based management, a critical component of sustainable development approaches.

Advanced EA implementations enabled comprehensive knowledge asset management, supporting research collaboration and innovation capacity. Ecuador’s hybrid EA-business infrastructure demonstrated how integrated architectures could preserve institutional knowledge while enabling cross-dimensional analysis for quality management and accreditation. ³¹

3.4.3 Alignment with global development goals

A key finding of this review is the consistent link between these EA-driven outcomes and broader development objectives. The reported benefits in efficiency, cost optimization, and access directly contribute to the United Nations’ 2030 Agenda for Sustainable Development. ³⁴ Table 3 summarizes the primary EA outcomes identified in the literature and maps their alignment with specific Sustainable Development Goal (SDG) targets, illustrating the role of EA as an enabler of wider socio-economic development.

The results section primarily provides information to address research questions RQ1 and RQ2, while the discussion part mainly focuses on interpreting RQ3 regarding transferability to fragile contexts and its application to Yemen. To synthesize the distribution of evidence across the three research questions, Table 4 presents the aggregated count of studies according to their primary RQ focus. Most studies contributed to more than one RQ, but the counts reflect the area of deepest analysis in each case. The complete Study-to-RQ Evidence Map, showing each study’s primary and secondary RQ coverage and associated thematic subsections, is provided in the supplementary matrix. ²³

4.1.1 Framework dominance and the value of adaptation

The review confirms TOGAF’s position as the de facto standard, utilized in 26 of 30 studies (86.7%) for its adaptability in complex academic settings. ⁹ Its Architecture Development Method (ADM) is especially valuable for fragile contexts, as its iterative nature allows institutions to build capacity incrementally, a critical feature where “big bang” transformations are likely to fail. ^{8, 27, 36} However, the finding that 23.3% of studies used hybrid approaches, such as Finland’s integration of TOGAF with ArchiMate and BPMN ¹⁰ or Egypt’s with COBIT5, ¹² implies that the most effective strategy for a context like Yemen is to blend international best practices with locally adapted models.

4.1.2 Implementation barriers and success factors

The data consistently points to a core set of challenges and enablers, summarized in Table 5. Resource constraints (financial and human) emerged as the most frequent barrier, cited in 27 studies (90%), ^{5, 37} followed by organizational resistance to change (76.7%) ^{1, 2} and the technical complexity of integrating legacy systems (70%). ^{27, 28} Correspondingly, the most cited success factors were strong executive sponsorship (80%), ^{15, 38} comprehensive stakeholder engagement (73.3%), ^{6, 39} and a phased, iterative implementation approach (86.7%). ^{7, 40} Small-scale pilots, recommended in 24 studies (80%), were identified as a key risk mitigation strategy to demonstrate value and reduce resistance before scaling. ^{6, 41}

4.2.1 Models for resource-constrained implementation

The findings offer clear, actionable models for fragile contexts. Libya’s Collaboratively Developed Enterprise Resource Planning (CD-ERP) model, which used a multi-tenant cloud architecture to achieve 40-60% cost reductions, demonstrates how consortium approaches can overcome individual capacity constraints. ⁵ Similarly, Thailand’s use of Agile Enterprise Architecture (AEA) provides a stakeholder-driven, iterative roadmap for managing complexity and building resilience in environments facing frequent change. ⁶ These cases demonstrate that, with the appropriate strategy, managing fragility becomes a challenge rather than a hindrance to progress. The use of sector-specific reference models like HERM further reduces development costs and risk by allowing institutions with low EA maturity to adopt proven frameworks incrementally. ³⁵

4.2.2 Connecting EA to broader development goals (ICT4D & SDGs)

The outcomes reported in Table 3 show that EA’s impact extends far beyond technical modernization. The efficiency gains (15-35%) and cost reductions (20-40%) translate directly into reallocating scarce resources toward core educational missions. ^{4, 31} This aligns EA with the principles of ICT4D and contributes to multiple Sustainable Development Goals (SDGs). For instance, enhanced digital infrastructure supports SDG 9 (Industry, Innovation, and Infrastructure), while efficiency gains support SDG 8 (Decent Work and Economic Growth). ¹⁰ Furthermore, by using EA to create unified authentication systems and mobile-friendly interfaces, institutions have demonstrably improved access for underrepresented groups, supporting SDG 5 (Gender Equality). ^{5, 32} In addition, EA’s role in enabling institutional resilience was highlighted in post-pandemic contexts, where institutions with mature EA frameworks transitioned to remote learning 40-50% faster. ³ This capacity for crisis response is a critical strategic asset for a conflict-affected context like Yemen.

4.2.3 Building institutional resilience and donor confidence

In post-pandemic contexts, institutions with mature EA frameworks transitioned to remote learning 40-50% faster, highlighting EA’s role in building institutional resilience against disruption. ³ This resilience, coupled with the improved capacity for evidence-based decision-making reported in 83% of studies, ^{31, 37} strengthens institutional legitimacy. This feature is critical for donor coordination, as EA frameworks provide the standardized project designs and impact measurement systems that build donor confidence. As an example, Colombia’s CHE2A demonstrated how a national reference architecture can better align government priorities with international development support. ¹¹ The emphasis on sustainability is paramount; Indonesia’s experience showed that investing in local EA competency led to 60% better long-term sustainability than relying on external consultants. ^{33, 42}

The convergence of these findings points toward a three-phase implementation pathway suited to Yemen’s specific circumstances. This pathway, detailed in Table 6, is designed to be incremental, collaborative, and capacity-building, directly addressing the challenges of institutional fragmentation and political instability. •

Phase 1: Foundation Building (Years 1-2): Focus on establishing foundational IT governance and developing a national higher education EA reference model adapted from HERM. This should be implemented through pilot projects at 3-4 accessible universities using a consortium model to share costs and build trust, as demonstrated in Libya ^{5, 35} ;

This pathway synthesizes lessons from the most relevant cases in our analysis, emphasizing collaborative governance with incremental progress and capacity-building approaches proven effective in resource-constrained environments.

This review contributes to Enterprise Architecture and Information Systems literatures by demonstrating how EA frameworks function as strategic organizational transformation tools in complex, multi-institutional environments. The evidence reveals that successful EA implementations in challenging organizational contexts require explicit integration of change management principles: stakeholder engagement, incremental adoption, capability development, and alignment with institutional governance structures.

The findings extend existing EA literature by discussing implementation dynamics in fragile organizational contexts, a significant gap in current scholarship that predominantly focuses on stable, well-resourced enterprise environments. The evidence demonstrates that while universal EA principles (such as those in TOGAF) remain relevant, their application requires substantial adaptation to address resource constraints, governance complexities, and institutional autonomy characteristic of federated higher education systems.

For the broader Information Systems literature, this research demonstrates how enterprise-wide digital transformation initiatives can simultaneously address technical interoperability challenges and organizational development objectives such as process optimization, governance improvement, and stakeholder coordination. The integration of EA frameworks with adaptive implementation strategies offers a pathway for achieving sustainable IT-enabled organizational change that extends beyond technology deployment to encompass institutional and systemic transformation. Theoretically, it positions EA within the ICT4D paradigm as a mechanism for building sustainable, resilient, and equitable digital ecosystems, contributing directly to long-term development outcomes as outlined in the SDGs.

Our findings also resonate with established frameworks beyond EA itself. The phased pathway proposed for Yemen aligns with EA maturity models, which advocate for an incremental progression from foundational to optimized states. ⁴³ Furthermore, the governance challenges identified, such as balancing central coordination with autonomy, are central concerns in IT governance frameworks like COBIT 2019, which emphasizes stakeholder needs, ⁴⁴ and ISO/IEC 38500, which assigns clear responsibilities for IT performance, strategy, and conformance. ⁴⁵ From an organizational change perspective, the documented barriers of resistance and lack of sponsorship reflect well-known challenges in models like Kotter's 8-Step Change Model, highlighting that EA implementation in fragile contexts is as much a change management challenge as a technical one. ⁴⁶

The study also contributes methodologically by providing a systematic, evidence-based approach to EA adoption in complex organizational settings, offering practitioners and researchers a replicable framework for evaluating and implementing EA in challenging institutional contexts.

Several limitations constrain the generalizability of these findings. The predominance of literature from relatively stable contexts may limit applicability to severely fragile environments, though the inclusion of studies from challenging contexts like Libya provides some mitigation. Publication bias toward successful implementations may underestimate the difficulty of EA adoption in fragile contexts, suggesting the need for more research documenting failures and partial successes.

Future research should examine the long-term sustainability of EA implementations in fragile contexts, particularly focusing on how initiatives adapt to changing political circumstances and donor priorities. Comparative studies examining EA adoption across different types of fragile states could provide more detailed knowledge about the contextual factors that influence success. Additionally, research examining the relationship between EA maturity and broader institutional resilience could inform both the EA and development literatures.

Future research should focus specifically on TOGAF’s flexibility mechanisms and develop context-specific adaptations for fragile state environments. A dedicated study examining TOGAF’s modular components, governance layers, and implementation phases could produce a simplified, customized TOGAF variant tailored for Yemen’s higher education sector, addressing resource constraints, capacity limitations, and stakeholder coordination challenges unique to post-conflict environments.

The evolving nature of digital technologies and their applications in education suggests the need for ongoing research examining how emerging technologies (artificial intelligence, blockchain, and the Internet of Things) can be integrated into EA frameworks designed for fragile contexts. Such research would inform both theoretical understanding and practical implementation of technology-enabled institutional transformation in challenging environments.