Influence of Early Architectural Decisions on Modular Platform Development

Abstract

Platform-based product development is a central strategy for dealing with complexity, en- abling reuse and supporting product variation within large-scale organizations, especially within original equipment manufacturing (OEM). Architectural decisions within early phases are, however, greatly affected by high uncertainty, leading to challenges within the organization. This thesis examines one organization in particular, researching ways of working and conditions needed to navigate uncertainty and improving decision-making during early phases of modular platform development. Distinctive for the study is a qualitative and iterative research where a literature study is combined with empirical data gathered from employees within the organization. The data gathered consist of contextual observations, a survey based on 25 respondents, as well as twelve expert interviews with selected employees within the organization. The empirical results show a lack of common definitions considering core concepts such as "architecture" and "platform," creating semantic noise and misunderstanding between departments. Furthermore, path dependency is identified where the first project tends to unintentionally dictate the long-term rules and limit the platform’s future flexibility. This is amplified by unclear decision mandates, a lack of "platform-first thinking," as well as short-term goals. To deal with these challenges, the thesis proposes a framework supporting decision-making in early phases structured around five key areas: (1) establishing a common language, (2) ensuring requirements-based definitions of common solutions, (3) conducting platform impact assessments, (4) clarifying cross-functional decision mandate and ownership, as well as (5) enforcing the traceability of decision rationales. The framework is designed as a flexible checklist giving the organization a structured method for dealing with uncertainty in early phases, supporting conscious decisions, while not compromising the platform’s long-term flexibility.