Design Science Research (DSR)

• Design Science Research (DSR) is a research paradigm that uses artifact creation and iterative evaluation to solve real-world problems.
• It emphasizes rigorous cycles of design, demonstration, and evaluation to ensure artifact validity and generalizability.
• DSR bridges practical relevance with scientific rigor by producing actionable models, methods, and instantiations for IT innovation.

Design Science Research (DSR) is a mature research paradigm central to information systems, software engineering, and allied disciplines, characterized by its dual commitment to solving real-world problems through the creation of innovative artifacts and producing rigorous, generalizable knowledge. DSR unites artifact-centric design with empirical investigation in iterative cycles, systematically supporting artifact conception, development, demonstration, evaluation, and theoretical abstraction (Jat et al., 28 Aug 2024, Peffers et al., 2020, Pastor et al., 13 Jul 2024, Kroop, 16 Feb 2025, Larsen et al., 12 Mar 2025). Its outputs span models, methods, constructs, and instantiations, consistently emphasizing specification, evaluation, and communication aligned to both practical relevance and scientific rigor.

1. Foundations and Core Constructs

DSR formalizes artifact-driven knowledge creation. Hevner et al.’s framework defines DSR as the creation and rigorous evaluation of artifacts—constructs, models, methods, or instantiations—that address concrete problems, underpinned by three core cycles: relevance (problem–context linkage), design (build–test refinement), and rigor (theoretical foundation and reproducibility) (Jat et al., 28 Aug 2024, Pastor et al., 13 Jul 2024, Peffers et al., 2020).

Artifacts are defined as designed entities—be they conceptual, procedural, or instantiated software systems—that effect change in a problem context. These are evaluated not just for technical correctness, but for purpose validity (do they solve the intended problem), design validity (are they usable and contextually fit), instrument validity (are evaluation and measurement methods adequate), technical validity (are artifacts robust), and generalizability (do results transfer across contexts) (Kroop, 16 Feb 2025).

The “technological rule” is at the core of DSR theory: "To achieve <Effect> in <Situation> apply <Intervention>." Each DSR project ideally results in the articulation, instantiation, and empirical validation of such rules, supporting both practical implementation and theoretical generalization (Engström et al., 2019).

Key pillars:

• Relevance: Anchoring research in genuine, industry-grounded problems.
• Rigor: Employing validated methods, established theories, and reproducible processes.
• Design: Centering cycles on build–evaluate feedback to drive artifact maturity.

2. Standard DSR Process Models

Most DSR frameworks instantiate a six-phase process:

1. Problem Identification and Motivation: Precise scoping of practical relevance using literature review, stakeholder interviews, and systematic gap analysis (Peffers et al., 2020, Pastor et al., 13 Jul 2024, Benali et al., 2021).
2. Definition of Solution Objectives: Translating problem dimensions into quantifiable or descriptive objectives and identifying success criteria (Pastor et al., 13 Jul 2024, Peffers et al., 2020, Benali et al., 2021).
3. Artifact Design and Development: Constructing models, prototypes, or processes that address the objectives, mapping kernel theories to actionable artifacts (Kroop, 16 Feb 2025, Dellermann et al., 2021).
4. Demonstration: Applying the artifact in a realistic or simulated context to verify feasibility (Pastor et al., 13 Jul 2024, Peffers et al., 2020, Ballandies et al., 2022).
5. Evaluation: Using empirical, analytical, or mixed methods to assess artifact performance on the specified metrics, employing benchmarks, case studies, field trials, statistical analyses, and qualitative feedback (Elahidoost et al., 11 Feb 2024, Kroop, 16 Feb 2025, Larsen et al., 12 Mar 2025).
6. Communication: Rigorously documenting and disseminating problem, artifact, process, evaluation, and findings to scientific and practitioner audiences (Peffers et al., 2020, Pastor et al., 13 Jul 2024, Benali et al., 2021).

Peffers et al. (Peffers et al., 2020) and Hevner et al. frameworks (Pastor et al., 13 Jul 2024, Kroop, 16 Feb 2025) dominate, sometimes augmented by integrated (e.g., Österle) or dual-cycle (Wieringa) models that emphasize either artifact creation (design cycle) or empirical investigation (empirical cycle) (Pastor et al., 13 Jul 2024).

Process models are inherently iterative, supporting feedback loops from demonstration/evaluation back to earlier phases for artifact refinement, as demonstrated in complex domain applications such as NLP-based requirements engineering (Elahidoost et al., 11 Feb 2024) and cloud-services configuration (Benali et al., 2021).

3. Artifact Types and Design Principles

Artifacts in DSR are classified along granular axes:

Artifact Type	Examples	Primary Objective
Constructs	Domain-specific vocabulary, ontologies	Enable specification and communication
Models	Conceptual models, design patterns, meta-models	Abstract system or phenomenon structure
Methods	Stepwise procedures, processes, evaluation checklists	Guide action and decision making
Instantiations	Prototypes, software tools, implemented systems	Realize concepts in operational form

Selection is context- and gap-dependent: conceptual gaps warrant constructs/models, procedural gaps are filled by methods, and practical gaps by instantiations. Most DSR projects produce composite artifacts (e.g., model+method+prototype) (Pastor et al., 13 Jul 2024, Peffers et al., 2020, Dellermann et al., 2021, Jat et al., 28 Aug 2024).

Principle design procedures, as evidenced in DSR exemplars, include:

• Iterative prototyping: Rapid development/testing to surface hidden assumptions and align stakeholder expectations (Elahidoost et al., 11 Feb 2024, Ballandies et al., 2022).
• Alignment with stakeholder values: Systematic mapping of value-sensitive requirements into design principles, as activated in blockchain-based customer feedback systems (Ballandies et al., 2022).
• Explicit traceability: Connecting requirements to atomic elements (sentences, clauses) or kernel theories, essential in correctness-driven domains (Elahidoost et al., 11 Feb 2024, Kroop, 16 Feb 2025).
• Integration of automation and human-in-the-loop: Ensuring solution correctness while adhering to domain-specific nuances (Elahidoost et al., 11 Feb 2024, Dellermann et al., 2021).

4. Evaluation Frameworks and Validity

The rigor of artifact evaluation is foundational. Classic frameworks (Hevner, Peffers, Österle) historically emphasized purpose validity and empirical evaluation but were limited in operationalizing instrument and design validity (Kroop, 16 Feb 2025).

Kroop (Kroop, 16 Feb 2025) synthesizes five essential artifact validity dimensions:

1. Instrument validity: Reliability and construct-alignment of measurement tools.
2. Technical validity: Correct, robust, and stable artifact operation.
3. Design validity: Usability, contextual fit, and aesthetic quality.
4. Purpose validity: Goal attainment and clear causal linkage between artifact and observed effects.
5. Generalization: Replicability and transferability across contexts.

Recent advances advocate a “multidimensional evaluation” explicitly incorporating all five validity types, mandating at least instrument and purpose validity in every project. Best practices include defining evaluation criteria up front, structuring validation procedures by type, and transparently documenting results for each validity axis (Larsen et al., 12 Mar 2025, Kroop, 16 Feb 2025).

The Design Science Validity Framework (Larsen et al., 12 Mar 2025) further categorizes claims and validity types, distinguishing criterion (artifact achieves utility), causal (particular feature causes effect), and context (utility generalizes) claims. Standardized mappings between claim and validity subtype, along with systematic reporting templates, are required for transparent evidence chains linking design to knowledge claims.

5. Application Domains and Case Study Patterns

DSR is prominent in domains where complex, ill-structured problems require not just new technology, but embodied domain knowledge, iterative stakeholder calibration, and system-level theorizing. Key application areas include:

• Requirements Variability Management: NLP-driven traceability pipelines, enhancing manual analysis and supporting configuration/test coverage optimization, via cycles of build–feedback–evaluation with practitioners (Elahidoost et al., 11 Feb 2024).
• Cloud-Services Configuration: Platforms integrating SPL theory, agent architectures, and runtime adaptation mechanisms, employing formal models and multi-criteria optimization frameworks (Benali et al., 2021).
• mHealth and Healthcare Informatics: Iterative design and deployment of mobile/clinical apps, validated via field trials employing usability indices, engagement metrics, and context-aware protocols (Jat et al., 28 Aug 2024).
• Abstractions for Distributed Scientific Applications: Pilot-job middleware and resource management frameworks, developed via pattern analysis, performance modeling, and deployments across heterogeneous clusters (Luckow et al., 2020).
• Blockchain-based Socio-Technical Systems: Value-sensitive, incentive-aligned customer feedback ecosystems evaluated using controlled field experiments and systematic stakeholder mapping (Ballandies et al., 2022).
• Hybrid-Intelligence Decision Support: Combining human crowd-sourcing and ML-based guidance for business-model validation, employing composite artifact models and two-level evaluation loops (Dellermann et al., 2021).
• Sustainability Requirements Engineering: Mechanisms for mapping UN SDGs to software/system requirements using ontologies, Delphi consensus, and structured metrics for coverage and impact (Brooks, 2020).

Case studies repeatedly demonstrate that DSR cycles require context-aware adaptation and robust feedback integration to meet both industrial and scientific targets (Elahidoost et al., 11 Feb 2024, Ballandies et al., 2022, Jat et al., 28 Aug 2024, Kroop, 16 Feb 2025).

6. Methodological Integration and Pedagogy

Recent literature emphasizes the need for robust teaching, adoption, and method transfer for both novice and advanced researchers. Cohesive educational approaches anchor on:

• Explicit artifact–context–problem identification: Mandatory in early research design phases (Pastor et al., 13 Jul 2024).
• Cycle planning and nested subcycles: Structured iterative refinement processes support artifact maturation and knowledge consolidation (Knauss, 2020).
• Research question structuring: Parallel alignment of problem, design, and evaluation RQs, with evidence tracking per iteration (Knauss, 2020).
• Role-driven guidelines: Differentiated best practices for students, supervisors, and industry mentors (Knauss, 2020).
• Integration of value-sensitive and stakeholder-centric methods: Direct mapping of values to requirements to restrict and focus the design space, improving efficiency (Ballandies et al., 2022).

Pilot surveys reveal most DSR adoption is supervisor-driven, and artifacts produced include models, prototypes, and methods. Exemplar-based instruction, detailed phase mapping, and early, iterative evaluation are prioritized (Pastor et al., 13 Jul 2024).

7. Challenges, Implications, and Future Directions

DSR faces ongoing methodological evolution, with several persistent and emergent challenges:

• Validity Gaps: Instrument and design validity remain underdeveloped in classic frameworks, risking acceptance of artifacts with weak measurement, poor usability, or limited generalizability (Kroop, 16 Feb 2025).
• Evaluation Complexity: Multimodal artifacts and heterogeneous deployment environments require tailored, multidimensional assessment protocols (Kroop, 16 Feb 2025, Larsen et al., 12 Mar 2025, Jat et al., 28 Aug 2024).
• Scalability and Generalizability: DSR projects often risk over-specialization to local context; formal external validity mechanisms are infrequently operationalized (Kroop, 16 Feb 2025, Larsen et al., 12 Mar 2025).
• Balance of Rigor and Relevance: Successful DSR must manage the tension between rigorous evaluation (peer standards) and practical adoption (stakeholder buy-in), often necessitating human-in-the-loop systems (Elahidoost et al., 11 Feb 2024, Ballandies et al., 2022).

Emerging recommendations include:

• Adopting prescriptive validity frameworks and making validity documentation mandatory (Kroop, 16 Feb 2025, Larsen et al., 12 Mar 2025).
• Integrated value-sensitive design pipelines to constrain design space efficiently (Ballandies et al., 2022).
• Systematic use of Q-methodology and participatory tools for surfacing stakeholder perspectives and improving design resonance (Nurhas et al., 2019).
• Modular process models supporting agile, continuous-delivery contexts, with rapid feedback and stakeholder validation (Knauss, 2020, Kroop, 16 Feb 2025).

Design Science Research remains an actively evolving paradigm, with best practices centered on explicit process structuring, multidimensional validity, and deep stakeholder integration, thereby enabling scientifically robust and actionable solutions for complex socio-technical problems.