OBJECTIVES: We aimed to identify when and how integration should take place within evaluations of complex population health interventions (PHIs). STUDY DESIGN: Descriptive analytical approach. METHODS: We draw on conceptual insights that emerged through (1) a working group on integration and (2) a diverse range of literature on case studies, small-n evaluations and mixed methods evaluation studies. RESULTS: We initially sought techniques to integrate analyses at the end of a complex PHI evaluation. However, this conceptualization of integration proved limiting. Instead, we found value in conceptualizing integration as a process that commences at the beginning of an evaluation and continues throughout. Many methods can be used for this type of integration, including process tracing, realist evaluation, congruence analysis, general elimination methodology/modus operandi, pattern matching and contribution analysis. Clearly signposting when integrative methods should commence within an evaluation should be of value to the PHI evaluation community, as well as to funders and related stakeholders. CONCLUSIONS: Rather than being a tool used at the end of an evaluation, we propose that integration is more usefully conceived as a process that commences at the start of an evaluation and continues throughout. To emphasize the importance of this timing, integration can be described as comprising 'Work Package Zero' within evaluations of complex PHIs.