Across many areas of Service Design a clear tension is found between a scientific approach to service improvement and its evaluation. Drawing from their respective healthcare service design experiences, one from a design background, the other from an organisational sociology perspective, the authors question the way ‘evidence’ is conceptualised and used to assess the relative effectiveness of new interventions within healthcare. In doing so, they challenge the domination of the quality improvement movement by a positivist paradigm with its plethora of scientific and technology-based solutions based on guidelines, scorecards, metrics and measurement systems. This paradigm often does not sit comfortably with the complexities of daily life, particularly within healthcare organisations where ‘proven’ innovations must become part of the routine practice of multiple teams comprising individuals with very different disciplinary backgrounds and hierarchical status. The authors argue that the implementation challenge is significantly shaped by less well-attended issues such as culture, language and cognition, identity and citizenship. This essay addresses the implementation ‘gap’, and argues that design-based and social science perspectives (with their common origins) can make a significant contribution to reconciling the science/art divide to the benefit of service improvement. It explores the positioning of designers in relation to working alongside or within healthcare service organisations and their consequent need to relate to different measurement systems and language to achieve legitimacy and see their approaches assimilated into routine practice. Finally, the issues and arguments arising from the healthcare context are discussed for their wider value and relevance to the field of Service Design as a whole.