As the COVID-19 pandemic has demonstrated, barriers to vaccination uptake are heterogeneous and vary according to the local context. We argue that a more systematic consideration of local social and behavioural mechanisms could improve the development, assessment and refinement of vaccination uptake interventions. The EBM+ approach to evidence appraisal, which is a development of a recent line of work on the epistemology of causality, provides a means to evaluate mechanistic studies and their role in assessing the effectiveness of an intervention. We argue that an EBM+ methodology offers several potential benefits for research on vaccination uptake interventions. It also motivates the use of detailed mechanistic models, rather than the high-level logic models used by process evaluations, for example.