Requirements Engineering for Low-Code Development (RE4LCD): An Empirical Study from Nordic and Central Europe

Abstract

Context: Low-code development platforms (LCDPs) combine visual development, reusable components, platform services, integration mechanisms, and opportunities for custom code. These characteristics challenge traditional requirements engineering (RE) because platform capabilities and constraints influence how requirements are elicited, negotiated, documented, validated, and realised. Objective: This study investigates how requirements engineering is performed in low-code development settings and how socio-technical configurations emerge when business stakeholders, developers, platform vendors, and platform capabilities jointly shape requirements work. It also examines how Levy and Hadar's Requirements Engineering for No-Code Development (RE4NCD) framework can be extended to the low-code context. Method: An interpretive qualitative study was conducted using fifteen semi-structured interviews with professionals from Nordic and Central European organisations experienced in low-code development. Drawing on Socio-Technical Systems theory, data were analysed using a Gioia-informed thematic analysis. The resulting RE4LCD framework was further assessed through retrospective mapping to the empirical findings and a limited face-validation exercise based on follow-up discussions with several available interview participants. Results: The analysis identified five dimensions of requirements engineering in low-code environments: requirements collection methods, legacy system handling, business and platform context understanding, quality requirements management, and platform-specific considerations. The findings show that requirements work becomes more visual, iterative, and demonstration-centred; legacy systems are selectively reinterpreted rather than replicated; quality responsibility is distributed between project teams, platform configuration, custom code, integrations, and vendors; and platform feasibility becomes an early requirements concern. The follow-up discussions highlighted the growing importance of AI-assisted functionalities in LCDPs, leading to the inclusion of AI feature awareness, AI expectation management, and AI-assisted component generation in the framework. LCDPs therefore function as socio-technical mediators rather than passive implementation tools. Conclusion: The study proposes the Requirements Engineering for Low-Code Development (RE4LCD) framework, extending RE4NCD to a more complex low-code context, where requirements may be realised through configuration, custom code, integration, workarounds, or rejection.