Architecting to Ensure Requirement Relevance: Keynote TwinPeaks Workshop

Research has shown that up to two thirds of features in software systems are hardly ever used or not even used at all. This represents a colossal waste of R&D resources and occurs across the industry. On the other hand, product management and many others work hard at interacting with customers, building business cases and prioritizing requirements. A fundamentally different approach to deciding what to build is required: requirements should be treated as hypothesis throughout the development process and constant feedback from users and systems in the field should be collected to dynamically reprioritize and change requirements. This requires architectural support beyond the current state of practice as continuous deployment, split testing and data collection need to be an integral part of the architecture. In this paper, we present a brief overview of our research and industry collaboration to address this challenge.