A comparative analysis of methodological adoption, adaptation and implementation of design science research (DSR) across engineering disciplines

Design Science Research has emerged as a prominent methodological paradigm for engineering disciplines seeking to bridge the gap between theoretical knowledge and practical innovation through systematic artefact creation (Hevner et al., 2004). Whilst DSR originated within information systems research, its principles of rigorous artefact development, iterative evaluation and relevance-driven design have diffused across diverse engineering domains (Peffers et al., 2007). However, the way different engineering disciplines adopt, adapt and implement DSR frameworks remains inadequately understood, limiting opportunities for cross-disciplinary learning and methodological refinement.

Engineering disciplines operate within markedly distinct epistemological traditions, face unique practical constraints and serve different stakeholder communities. Software engineering emphasises empirical validation and user experience; manufacturing engineering prioritises operational efficiency, quality management and statistical process control; construction and environmental engineering navigate project-based complexity and long-term socio-technical systems; biomedical, mechanical and aerospace engineering address safety-critical regulatory requirements alongside innovation demands. These characteristics suggest that DSR adoption will exhibit systematic disciplinary variations worthy of comparative investigation. Canonical DSR scholarship has established robust foundations: Hevner et al. (2004) formalised seven guidelines emphasising relevance, rigour and artefact utility; Peffers et al. (2007) operationalised these through their six-phase Design Science Research Methodology (DSRM); and Gregor and Hevner (2013) positioned DSR outputs along dimensions of problem and solution maturity, enabling researchers to articulate theoretical value beyond specific implementations. Iivari (2015) further observed that disciplinary context, artefact type and knowledge objectives shape methodological choices in ways canonical frameworks do not fully accommodate. This body of scholarship has advanced DSR predominantly within information systems contexts; no comparative study has systematically examined how different engineering disciplines select, adapt and implement DSR frameworks, nor whether universal success principles operate consistently across their distinct epistemological traditions.

Despite growing DSR adoption across engineering domains, three significant research gaps persist. First, the literature lacks systematic cross-disciplinary comparison of framework selection and adaptation: extant studies focus on single disciplines (Engström et al., 2020; Weber, 2023), precluding identification of patterns or the conditions under which canonical frameworks suffice versus require contextual adaptation. Second, no synthesis distinguishes universal success principles from discipline-specific adaptations. Third, cross-disciplinary methodological transfer opportunities remain unexplored.

Three research questions guide the investigation:

This study makes three contributions. Theoretically, it provides the first cross-disciplinary empirical synthesis of DSR adoption patterns, advancing understanding of how domain complexity and socio-technical context shape framework selection, a relationship theorised but not empirically demonstrated in prior work (Drechsler and Hevner, 2016; Weber, 2023). Methodologically, comparative systematic review demonstrably identifies both universal principles and contextual variations in research methodology adoption, offering a transferable analytical model beyond DSR. Practically, the study equips engineering researchers with evidence-based guidance for framework selection, evaluation design and challenge mitigation.

The remainder of the paper is structured as follows.: Section 2 reviews DSR foundations, framework evolution and disciplinary applications. Section 3 describes the systematic review methodology. Section 4 presents comparative findings across disciplines. Section 5 discusses theoretical contributions, practical implications and limitations. Section 6 concludes.

Design Science Research emerged as a distinct research paradigm addressing the creation and evaluation of innovative artefacts intended to solve identified organisational problems (Simon, 1996; Hevner et al., 2004). Unlike explanatory science seeking to understand existing phenomena, design science aims to create new, useful artefacts through rigorous, iterative development and evaluation processes. The paradigm gained prominence in information systems through Hevner et al.’s (2004) articulation of seven guidelines emphasising relevance, rigour, design as search, research contributions, evaluation, communication and environment understanding.

Peffers et al. (2007) advanced DSR operationalisation through their Design Science Research Methodology (DSRM), specifying six sequential activities: problem identification and motivation, objectives definition, design and development, demonstration, evaluation and communication. This process model provided researchers with structured guidance whilst maintaining flexibility for iteration and adaptation. The DSRM framework gained widespread adoption, particularly in information systems and software engineering, establishing a canonical reference point for DSR implementation.

March and Smith (1995) contributed foundational thinking regarding DSR outputs, distinguishing four artefact types: constructs (vocabulary and symbols), models (abstractions and representations), methods (algorithms and practices) and instantiations (implemented systems). This typology helped researchers articulate their knowledge contributions beyond specific implementations, emphasising generalisable design knowledge. Gregor and Hevner (2013) later extended this thinking through their knowledge contribution framework, positioning DSR outputs along dimensions of problem maturity and solution maturity, ranging from invention (new problems and solutions) to routine design (known problems and solutions).

Evaluation constitutes a central DSR concern, with researchers debating appropriate methods for assessing artefact utility, quality and efficacy. Venable et al. (2016) proposed the Framework for Evaluation in Design Science (FEDS), distinguishing formative versus summative evaluation and naturalistic versus artificial evaluation contexts. This framework acknowledged that evaluation strategies should align with research goals, artefact maturity and available resources. Prat et al. (2015) developed a comprehensive taxonomy of evaluation methods for information systems artefacts, providing researchers with structured guidance for evaluation design.

As DSR diffused beyond information systems, debates emerged regarding framework universality, adaptation requirements and integration with domain-specific methodologies. Critics argued that canonical DSR frameworks, developed primarily within information systems contexts, might inadequately address the distinctive characteristics of other engineering domains, including different artefact types, validation requirements and stakeholder expectations (Iivari, 2015). These concerns prompted development of domain-specific DSR variants and hybrid approaches combining DSR with established engineering methodologies.

The concept of search in design science has gained prominence, particularly for complex, novel problems in which problem and solution spaces co-evolve (Tuunanen et al., 2024). Traditional DSR frameworks assume relatively stable problem definitions, but research addressing wicked problems or emerging domains may require exploratory search processes in which problem understanding and solution development proceed iteratively. This recognition has led to proposals for search-focused DSR variants emphasising discovery and exploration alongside solution development (Tuunanen et al., 2024).

Evaluation methodology constitutes another area of ongoing debate and development. Hevner et al. (2004) emphasised evaluation centrality, questions persist regarding appropriate evaluation methods for different artefact types, research stages and disciplinary contexts. Prat et al. (2015) developed a taxonomy of evaluation methods for information systems artefacts, distinguishing functional testing, structural evaluation and usability assessment approaches. However, the applicability of these methods across diverse engineering disciplines remains underexplored, with different domains potentially requiring distinct validation approaches aligned with disciplinary norms and stakeholder expectations.

The integration of DSR with other research methodologies has emerged as a significant trend, particularly in applied engineering contexts. Hybrid approaches combining DSR with action research, case study methods, ethnography and implementation science have been proposed to address limitations of pure DSR approaches and leverage complementary methodological strengths (Castro et al., 2025; Gough et al., 2024). These integrations raise questions about methodological coherence and the boundaries of DSR as a distinct paradigm.

A further area of methodological debate concerns the relationship between artefact development and organisational intervention. Sein et al. (2011) advanced Action Design Research (ADR) as a methodological response to canonical DSR's perceived separation of building and evaluation phases, arguing that constructing the IT artefact, intervening in organisational practice and evaluating outcomes are inseparable and mutually informing activities. This reconceptualisation is particularly relevant for understanding why high-complexity domains develop custom frameworks in which problem and solution spaces co-evolve, sequential DSR phases inadequately capture the emergent, stakeholder-embedded nature of artefact development (Miah and Genemo, 2016). Wall et al.’s (1992) foundational work on Information Systems Design Theory (ISDT), as interpreted by Gregor and Hevner (2013), further establishes that valid design knowledge must articulate meta-requirements, meta-designs and testable design hypotheses, criteria that engineering disciplines with different epistemological traditions operationalise distinctively. A key theoretical tension thus emerges between the linearised, phase-based progression assumed by canonical frameworks and the co-evolutionary, adaptive processes demanded by complex engineering domains. This tension is examined empirically across six disciplines in the present study.

Software engineering represents the most mature domain for DSR adoption, with extensive literature examining alignment between software development practices and design science principles (Engström et al., 2020). The discipline predominantly employs Peffers et al.’s (2007) DSRM integrated with Agile or Spiral lifecycle models, emphasising technological rules as theoretical contributions and generalisable design knowledge beyond specific implementations (Stickland et al., 2024). Evaluation combines prototype testing, expert reviews and user studies, though a productive tension persists between software engineering's empirical validation tradition and DSR's focus on artefact utility and design knowledge contribution (Apiola and Sutinen, 2021).

Manufacturing engineering applications of DSR frequently focus on developing maturity models, interoperability frameworks and process improvement, developing domain-specific variants tailored to production system characteristics rather than adopting canonical frameworks (Kirmizi et al., 2022; Stary et al., 2019). These adaptations reflect manufacturing's emphasis on operational efficiency, quality management and integration across complex value chains, contexts in which standard DSRM phases cannot accommodate the statistical rigour and stakeholder governance requirements of production system change. Evaluation employs iterative expert panels, intra-class correlation coefficients and Wilcoxon consensus tests alongside pilot implementations, reflecting the high stakes of production system deployment and the discipline's practitioner-oriented culture.

Construction engineering DSR applications address coordination challenges, project management inefficiencies and digitalisation opportunities through development of decision support tools and management dashboards (Gledson et al., 2023). The discipline's project-based nature and fragmented industry structure shape DSR adoption, with problem definition established through extensive stakeholder consultation prior to solution development. Evaluation prioritises practitioner feedback and prototype demonstration in authentic project contexts rather than controlled experimentation, reflecting the practical difficulty of experimental control in complex construction environments; key implementation challenges include workflow integration, data quality management and adoption across fragmented organisational structures.

Healthcare engineering DSR addresses clinical decision support, medical device development and health information systems through artefacts ranging from AI-based diagnostic tools to clinical workflow frameworks (Gough et al., 2024). The domain's complexity, regulatory requirements and patient safety imperatives necessitate distinctive DSR adaptations: healthcare researchers increasingly integrate DSR with implementation science, combining artefact development with systematic attention to adoption barriers and contextual factors. Co-evolution of problem and solution spaces is particularly prominent, requiring exploratory research processes in which problem understanding and solution development proceed simultaneously (Tuunanen et al., 2024). Staged validation progressing from laboratory to clinical settings is characteristic, shaped by regulatory expectations and ethical requirements.

Environmental engineering applications of DSR focus on developing models, decision support tools and frameworks bridging scientific knowledge and practical environmental management (Zare et al., 2024). The discipline's emphasis on socio-technical systems and stakeholder engagement shapes a distinctive DSR orientation: researchers advocate adaptations emphasising relevance cycles and rapid translation to practice and increasingly embed FAIR (Findable, Accessible, Interoperable, Reusable) principles to ensure knowledge transferability and sustained practitioner uptake. Evaluation emphasises stakeholder engagement, case study validation and transferability assessment across diverse contexts, with particular challenges in integrating diverse knowledge sources, managing uncertainty and demonstrating impact within the extended timeframes characteristic of environmental processes.

Mechanical and aerospace engineering DSR applications address product development, system design and interdisciplinary innovation challenges, with researchers integrating DSR with Spiral approaches and stage-gate processes that reflect the domain's risk management culture and regulatory requirements (Gujarathi et al., 2025; Viviani et al., 2024). The concept of echeloned DSR (eDSR), which decomposes complex projects into hierarchical sub-projects to manage system interdependencies and coordinate interdisciplinary teams, has emerged as a distinctive methodological innovation in this domain (Tuunanen et al., 2024). Evaluation employs staged testing, prototype demonstration and field trials aligned with engineering development cycles, with key challenges in managing technical complexity and balancing innovation against safety and reliability requirements.

Table 1 consolidates the principal success factors reported or demonstrated across the six disciplinary applications reviewed above, providing a cross-disciplinary baseline that Section 4 findings subsequently interrogate.

Despite growing DSR adoption across engineering disciplines, three gaps limit cross-disciplinary understanding and practice advancement. First, no comparative study examines how different engineering disciplines select and adapt DSR frameworks. Existing studies focus on single disciplines, preventing identification of cross-disciplinary patterns and contextual variations. Second, whilst individual studies report success factors and challenges, no synthesis distinguishes universal principles from discipline-specific adaptations. Third, opportunities for methodological innovation transfer across disciplines remain unexplored. This study addresses these gaps through comparative analysis of DSR adoption across six engineering disciplines, examining framework selection, artefact types, evaluation methods, success factors and implementation challenges.

This study employs a systematic literature review methodology to compare DSR adoption, adaptation and implementation across six engineering disciplines. The systematic review approach follows PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines (Page et al., 2021) to ensure transparent, reproducible identification and analysis of relevant literature, consistent with recent systematic reviews in engineering contexts that have similarly employed PRISMA-guided protocols (Anane et al., 2023; Atencio et al., 2022). The review encompasses four main phases: planning, searching, screening and data extraction and analysis.

Six engineering disciplines were selected based on three criteria: (1) established and documented DSR adoption in peer-reviewed literature, ensuring sufficient retrievable evidence for comparative analysis; (2) collective coverage of both physical and digital artefact types, enabling examination of artefact-type variation as a potential moderator of DSR adoption; and (3) representation of diverse epistemological traditions, from the formal verification culture of software engineering to the socio-technical complexity of environmental engineering. Whilst other engineering domains including mining, aeronautical, robotic, data and mechatronic engineering apply DSR in varying degrees, their exclusion reflects the absence of sufficient empirical DSR literature to support systematic comparison at the point of data collection. These excluded disciplines are acknowledged as a study limitation and constitute a productive direction for future research.

The literature search was conducted between October and November 2025 across seven electronic databases: IEEE Xplore, ScienceDirect, Scopus, Web of Science, ACM Digital Library, Engineering Village and SpringerLink. These databases were selected for comprehensive coverage across engineering disciplines, with deduplication procedures applied to address potential overlap.

Search terms combined DSR-related keywords with discipline-specific terms using Boolean operators:

As illustrated in Figure 1, the initial search retrieved 406 papers meeting basic relevance criteria. Title and abstract screening reduced this to 248 papers warranting full-text review. Papers were included if they: (1) employed DSR or related design-oriented methodologies, (2) addressed engineering contexts, (3) described framework adoption and adaptation and (4) provided sufficient methodological detail. During screening, 158 records were excluded for not meeting DSR criteria, not being engineering-focused, or not being empirical studies, consistent with PRISMA 2020 exclusion reporting conventions (Moher et al., 2009; Page et al., 2021).

Following title and abstract screening, 248 full-text articles were assessed for eligibility. Inclusion criteria required empirical DSR studies conducted in engineering disciplines with framework discussion and peer-reviewed publication between 2015 and 2025. During full-text assessment, 218 articles were excluded: 98 due to insufficient methodological detail, 67 for lack of framework discussion, 31 duplicates and 22 due to quality issues. This process resulted in 30 studies included in the qualitative synthesis, with balanced discipline distribution: Software/Computer (6), Manufacturing/Industrial (5), Construction/Civil (4), Biomedical/Healthcare (5), Environmental/Socio-Environmental (5) and Mechanical/Aerospace (5). Data extracted focused on framework adoption, artefact types, evaluation methods, success factors, implementation challenges and methodological innovations. Whilst five papers per discipline is modest, this corpus size is consistent with purposive sampling norms in qualitative evidence synthesis, in which prioritisation of information-rich cases that closely address the synthesis objective takes precedence over breadth of coverage (Booth et al., 2019; Palinkas et al., 2015); the analytical focus on framework adoption rationale, evaluation strategy and implementation challenges mitigates representativeness concerns (Paré et al., 2015).

Coding was conducted by two independent researchers, with inter-rater reliability exceeding 85%, above the 80% threshold recommended for reliable qualitative coding (O'Connor and Joffe, 2020). Sample adequacy is acknowledged as a limitation in Section 5.3 Data extraction employed a structured template capturing key information from each paper:

1. Framework selection and adaptation rationale

2. Artefact types developed

3. Evaluation methods employed

4. Success factors reported

5. Implementation challenges encountered

6. Methodological innovations introduced

Each paper was systematically coded according to these dimensions, enabling both within-discipline and cross-discipline comparative analysis. Coding was conducted iteratively, with initial categories refined through constant comparison as patterns emerged. Initial open coding produced 47 codes across the six extraction dimensions; through iterative consolidation and axial coding, these were reduced to 23 final categories distributed across framework adoption (6 categories), artefact types (5), evaluation methods (5), success factors (4) and implementation challenges (3).

Artefact type categories were derived deductively from March and Smith's (1995) foundational typology of constructs, models, methods and instantiations, extended inductively to accommodate domain-specific artefact forms identified in the corpus, including maturity models, decision support tools and physical prototypes, which represent engineering-specific elaborations of the instantiation category not fully captured in the original IS typology.

Critical success factors were identified through a two-stage process. In the first stage, papers were coded for explicitly reported success factors, defined as conditions or practices the authors attributed to positive DSR outcomes and for implicitly demonstrated factors, defined as methodological choices consistently associated with successful artefact development across multiple papers. In the second stage, factors appearing across three or more disciplines were classified as universal; those present in one or two disciplines were classified as discipline specific. This process produced five universal success factors and multiple discipline-specific adaptations discussed in Section 4.4 and Figure 3 panel (a).

For comparative analysis, qualitative intensity scores (1–10 scale) were assigned to capture the prevalence and emphasis of specific patterns within each discipline's literature. Scores were based on three criteria: (1) frequency of mention across papers within each discipline, (2) depth of discussion and emphasis given by authors and (3) reported implementation levels or adoption prominence. For the purposes of this study, adoption prominence refers to the relative frequency and prominence with which a given framework, artefact type, or evaluation method appears across the reviewed papers within a discipline, whilst implementation level denotes the depth of application reported, ranging from superficial reference to full methodological integration. Scores of 8–10 indicate high intensity/prevalence, 5–7 moderate and 1–4 low.

Both researchers independently scored patterns for each discipline, with final scores determined through consensus where initial scores differed. This qualitative scoring approach enabled systematic comparison across disciplines whilst acknowledging the interpretive nature of synthesising diverse literature. The scoring framework was applied consistently across all reviewed papers.

Analysis proceeded through three stages. First, within-discipline analysis examined consistency and variation in DSR adoption within each engineering domain. Second, cross-discipline comparison identified patterns transcending individual disciplines alongside discipline-specific characteristics. Third, thematic synthesis aggregated findings regarding universal success factors, common challenges and cross-disciplinary transfer opportunities.

Quality assessment employed criteria adapted from systematic review standards: (1) methodological rigour, (2) clarity of framework application, (3) comprehensiveness of evaluation and (4) contribution to DSR knowledge. Each paper was assessed against these criteria during full-text review, with papers demonstrating deficiencies across multiple dimensions excluded from analysis. All 30 included papers demonstrated adequate quality across these dimensions, though variability existed in evaluation comprehensiveness and theoretical contribution articulation.

DSR framework adoption exhibits significant disciplinary variation, reflecting differences in domain complexity, stakeholder characteristics and methodological traditions. Figure 2, panel (a) details adoption prominence across four framework categories for all six disciplines. Software engineering demonstrates the highest standardisation through consistent DSRM preference, aligned with established software development lifecycle models (Stickland et al., 2024). In contrast, manufacturing, environmental and mechanical engineering develop predominantly custom or hybrid frameworks addressing domain-specific constraints. Construction engineering shows a balanced profile, combining problem-focused DSRM with practitioner-collaborative approaches, whilst healthcare engineering integrates DSR with implementation science through adaptive, co-evolutionary frameworks (Gough et al., 2024). An inverse relationship is evident across Figure 2 panel (a) between framework standardisation and domain complexity: disciplines managing complex socio-technical systems develop more customised frameworks, whilst software engineering benefits from established process alignment. This pattern motivates the theoretical explanation developed in Section 5.1.

The artefact type taxonomy employed here is derived from March and Smith (1995) and refined through inductive coding as described in Section 3.3. Artefact type distribution across disciplines, detailed in Figure 2 panel (b), reflects both universal knowledge contribution needs and discipline-specific specialisations. Decision support tools show the broadest cross-disciplinary adoption, indicating a universal need for evidence-based decision-making in complex engineering contexts. Manufacturing and environmental engineering demonstrate the highest artefact diversity, developing simultaneously across models, frameworks, maturity models and decision support tools, reflecting multi-modal knowledge contribution strategies. Software engineering shows more focused development centred on software systems, whilst mechanical and aerospace engineering balances prototypes with decision support tools, consistent with the discipline's emphasis on physical demonstration and performance validation (Gujarathi et al., 2025). Manufacturing's distinctive emphasis on maturity models reflects its focus on digital transformation assessment and capability benchmarking (Kirmizi et al., 2022).

Evaluation method preferences, illustrated in Figure 2 panel (c), exhibit variation aligned with domain validation norms, stakeholder expectations and artefact characteristics. Three cross-disciplinary patterns are notable. First, manufacturing demonstrates the most rigorous evaluation profile, combining expert review, statistical validation and case studies in triangulated strategies using intra-class correlation coefficients and multiple consensus rounds (Kirmizi et al., 2022). Second, safety-critical disciplines, notably biomedical and aerospace engineering, demonstrate the highest overall evaluation intensity, reflecting regulatory expectations and risk management imperatives. Third, practice-oriented disciplines, including construction and environmental engineering, prioritise case studies and practitioner feedback over controlled experimentation, consistent with their project-based operational contexts (Gledson et al., 2023; Zare et al., 2024). These divergent profiles suggest that evaluation rigour should scale with deployment stakes rather than follow uniform standards across disciplines.

Figure 3, panel (a) illustrates critical success factors and challenges for DSR implementation across disciplines, revealing both universal principles and contextual emphases. Stakeholder engagement is the dominant universal success factor (8–9/10 across all disciplines), challenging assumptions about technical rigour primacy and confirming that relevance and practical impact depend fundamentally on practitioner involvement from research inception. Iterative design is universally recognised (7–9/10), whilst expert validation and clear benchmarks are most critical in manufacturing and mechanical/aerospace engineering, where high deployment stakes demand objective evaluation baselines. Mixed methods are broadly adopted (6–8/10), reflecting recognition that single evaluation approaches inadequately address the multiple quality dimensions of complex engineering artefacts.

Challenge profiles reflect disciplinary context and artefact characteristics. Complexity management is the most severe challenge in biomedical and mechanical/aerospace engineering (both 9/10), consistent with the multi-level, socio-technical nature of healthcare systems and safety-critical product development. Practitioner uptake is the primary barrier in environmental and construction engineering (both 8/10), reflecting gaps between academic outputs and practitioner needs in fragmented industry structures. Evaluation rigour presents a persistent cross-disciplinary challenge (5–8/10 across all disciplines), suggesting the need for improved DSR evaluation frameworks that address diverse disciplinary contexts.

Figure 3 panel (b) presents an integrated assessment of eight DSR characteristics, revealing systematic trade-offs between standardisation and adaptation, rigour and relevance and artefact diversity and focus. Framework standardisation varies most markedly across disciplines (2–7/10), confirming the inverse relationship with domain complexity identified in Section 4.1. Stakeholder engagement achieves the most consistent profile (8–9/10 across all disciplines), validating DSR's fundamental commitment to relevance. Manufacturing, biomedical and environmental engineering demonstrate the most balanced DSR maturity profiles, scoring highly across multiple characteristics simultaneously, whilst software engineering shows comparatively focused development concentrated in framework standardisation and iterative approach dimensions.

Phase adaptation patterns, illustrated in Figure 3 panel (c), reveal systematic differences in where disciplines invest methodological effort across the six DSR phases. Practice-oriented disciplines, notably construction and environmental engineering, invest most heavily in problem identification and demonstration phases, reflecting emphasis on genuine practice needs and tangible deployment evidence (Zare et al., 2024; Gledson et al., 2023). Research-intensive disciplines, including software, manufacturing and biomedical engineering, concentrate adaptation in design-development and evaluation phases, reflecting their commitment to technical rigour and multi-round validation. Balanced adaptation across all phases, observed most clearly in biomedical and mechanical engineering, is indicative of DSR methodological maturity and aligns with these disciplines' success factor profiles identified in Figure 2 panel (a).

The theoretical contributions presented in this section were derived through a three-stage process: first, cross-disciplinary patterns were identified by comparing adoption prominence scores across disciplines; second, patterns were tested against the prior literature success factors in Table 1 to assess corroboration or divergence; third, theoretical mechanisms were constructed to explain systematic patterns not attributable to sampling artefacts, grounded in DSR theory and complexity literature.

This study makes three theoretical contributions. First, it reveals systematic patterns in DSR framework adaptation related to domain complexity and sociotechnical integration requirements. High-complexity domains instantiate what Rittel and Webber (1973) term wicked problems, characterised by co-evolving problem-solution spaces and emergent requirements that canonical DSRM cannot accommodate, given its presupposition of stable problem definitions and sequential phase progression (Miah and Genemo, 2016). Additionally, high-complexity domains feature diverse stakeholder groups, evolving regulatory constraints and socio-technical interdependencies that cannot be fully specified a priori (Drechsler and Hevner, 2016), rendering standardised frameworks inadequate for encoding domain-specific governance and regulatory alignment processes. Researchers consequently develop bespoke, context-sensitive frameworks prioritising iterative search and stakeholder co-design, as Zare et al. (2024) advocate for socio-environmental contexts and as Sein et al. (2011) ground theoretically in ADR's reconceptualisation of concurrent building, intervening and evaluating. This pattern is corroborated by Engström et al. (2020) and Weber (2023), who document standardised DSR reporting in software engineering and argue that minimal rather than full standardisation is appropriate given disciplinary diversity; Stary et al. (2019), Goecks et al. (2019) and Coetzee (2019) independently confirm domain-specific adaptations in manufacturing and the universality of mixed evaluation approaches in human-centred engineering contexts.

Second, the identification of five universal success factors (stakeholder engagement, iterative design, mixed methods, expert validation, clear benchmarks) provides empirical validation of core DSR principles whilst acknowledging implementation variations. Stakeholder engagement's dominance (8–9/10 across all disciplines) challenges assumptions about technical rigour primacy, suggesting that relevance and practical impact depend fundamentally on practitioner involvement from research inception. Whilst these factors demonstrate universal applicability, their effectiveness is subject to contextual boundary conditions practitioners must recognise. Stakeholder engagement is constrained in community-oriented DSR contexts featuring high participant diversity and permeable organisational boundaries (Renzel et al., 2015). Iterative design is limited by resource and timeline constraints, a concern especially acute in construction and biomedical engineering (Miah and Genemo, 2016). Mixed evaluation methods remain insufficient when evaluations lack realistic end-user involvement (De Sordi et al., 2020). Expert validation can fail to capture usability issues when consensus is achieved on theoretical criteria without agreement on practical implementation (Moutinho et al., 2023). Clear benchmarks embedded in maturity models risk imposing weak theoretical foundations and false certainty regarding deterministic improvement pathways (Mettler, 2009). Awareness of these conditions enables researchers to anticipate contexts in which each factor requires supplementary mechanisms.

Third, the mapping of methodological innovations with cross-disciplinary transfer potential advances DSR methodology. Echeloned DSR's hierarchical decomposition (Mechanical/Aerospace) offers valuable approach for managing complexity in all engineering domains; healthcare's integration of DSR with implementation science provides a replicable model for systematically addressing adoption barriers; and environmental engineering's FAIR principles enhance artefact transferability across practitioner communities. These findings are independently corroborated by Engström et al. (2020), Weber (2023), Stary et al. (2019), Goecks et al. (2019) and Coetzee (2019), whose cross-disciplinary observations align with the disciplinary patterns identified here.

The findings provide actionable guidance for multiple stakeholder groups. For engineering researchers selecting DSR frameworks, the analysis supports a contingency approach grounded in domain complexity: standardised DSRM is appropriate for software projects with well-defined technical requirements and shorter development cycles, whilst complex socio-technical projects, notably those in healthcare, environmental and manufacturing engineering, benefit from custom frameworks explicitly accommodating stakeholder governance, regulatory alignment and iterative search processes, consistent with Weber's (2023) argument for minimal common schemas. Table 2 operationalises this guidance by mapping project characteristics to recommended approaches across the six disciplines examined.

For PhD students and early-career researchers, the study offers realistic expectations regarding DSR implementation challenges and success factors; awareness of the boundary conditions in Section 5.1 enables proactive mitigation strategies rather than reactive adaptation during fieldwork. For research methods educators, the comparative analysis provides pedagogically rich material illustrating how universal principles manifest distinctively across disciplinary settings, informing curriculum design that encompasses both canonical frameworks and domain-specific adaptations across construction, manufacturing and biomedical programmes.

From a sustainability perspective, environmental and manufacturing DSR artefacts, including whole-life-cycle assessment models and circular economy decision support tools, directly address sustainable materials design and embodied carbon reduction, with FAIR principles providing a transferable model for cross-community knowledge reuse (Zare et al., 2024). For journal editors and reviewers, the findings enhance understanding of legitimate DSR variation, potentially reducing inappropriate rejection of domain-appropriate evaluation approaches. DSR artefacts in healthcare, environmental and construction engineering carry substantive societal significance, delivering measurable improvements to patient safety, ecological sustainability and built environment outcomes respectively.

Several limitations warrant acknowledgement. First, the review focuses on English-language publications from selected databases, potentially excluding regional variations and non-English contributions; future research should examine DSR adoption in non-Western contexts and incorporate multilingual literature. Second, the 30-paper corpus, whilst providing analytical depth within each discipline, limits breadth of coverage; larger-scale bibliometric or text-mining analysis could reveal additional sub-disciplinary variations. Third, publication bias towards successful DSR implementations may overestimate success factor prevalence and underestimate implementation challenges; investigation of failed or abandoned projects would provide a more balanced account. Fourth, disciplinary boundaries are necessarily artificial, with many papers spanning multiple domains; sub-disciplinary analysis would refine the patterns identified here.

Fifth, the selection of six disciplines excludes emerging domains including mining, aeronautical, robotic, data and mechatronic engineering, in which DSR adoption may be growing but is not yet sufficiently documented for systematic comparison; these constitute productive directions for future research. Sixth, explicit success factor reporting is inconsistent across the reviewed literature; whilst Kirmizi et al. (2022) formally document success factors, most papers demonstrate them implicitly through methodological choices, introducing interpretive subjectivity that dual independent coding mitigates but cannot eliminate. Seventh, the qualitative intensity scoring approach (1–10 scale) constitutes a recognised limitation of this study. Whilst inter-rater reliability and explicit scoring criteria enhance consistency, alternative analytical approaches including frequency analysis, word co-occurrence mapping, or thematic network analysis would offer greater methodological transparency and are recommended for future cross-disciplinary DSR synthesis studies.

DSR methodology research has long acknowledged disciplinary variation in artefact-oriented research without systematically characterising it; this study resolves that gap through comparative analysis across six engineering disciplines, establishing that both universal principles and discipline-specific adaptations are empirically verifiable features of DSR practice rather than untested assumptions. The study's theoretical contribution lies not in identifying that disciplinary variation exists, but in demonstrating that it follows a systematic pattern explicable through domain complexity and socio-technical context. This provides the DSR research community with an explanatory framework, rather than merely a descriptive taxonomy, from which future theory development can proceed. The five universal success factors, together with their discipline-specific boundary conditions documented here for the first time in cross-disciplinary synthesis, shift the field's understanding from universal prescriptions towards context-aware principles, enabling more analytically grounded evaluation of DSR quality.

The methodological contribution establishes that cross-disciplinary comparative systematic review, structured around multi-dimensional extraction and independent dual coding, is a viable and replicable approach for generating comparative knowledge about research methodology itself, opening a pathway applicable to any tradition in which disciplinary variation is suspected but unexplored.

In practice, the contingency framework established here enables engineering researchers to make more defensible, evidence-based methodological choices aligned to their domain's epistemological and operational characteristics, rather than defaulting to canonical frameworks regardless of context. It also provides institutional stakeholders, including journal editors, programme designers and research funders, with an evidence base for evaluating and supporting DSR-based contributions more fairly and consistently across disciplines.

Future research, building on the limitations identified in Section 5.3, should prioritise three directions: longitudinal tracking of DSR adoption to test whether the patterns identified here are stable or responsive to methodological trends; investigation of failed DSR projects to balance the current literature's success bias; and extension of the comparative framework to excluded and emerging engineering domains. Together, these would transform the current cross-sectional picture into a dynamic account of DSR methodology evolution.

Engineering research needs methodological frameworks that are both rigorous and responsive to disciplinary context. This study affirms that DSR, properly adapted, satisfies both criteria and that the diversity of its implementation across engineering disciplines reflects methodological vitality rather than inconsistency.

This research was conducted in accordance with ethical principles and received approval from the University of Salford Ethics Committee.