In this talk, I hope to convince you that knowledge of mechanisms is important for the design, justification and evaluation of behavioral public policies (BPP), and I will give some suggestions how this knowledge can be acquired and applied. I start by discussing the concept of mechanisms, and how it pertains to BPP. I then offer four arguments for the importance of BPP mechanisms: they facilitate BPP design, justify BPP extrapolation, structure the aggregation of evidence for BPP effectiveness, and are often required for the normative evaluation of BPP in terms of welfare, autonomy and fairness. I then discuss to what extent such mechanistic knowledge is currently available, and what strategies would make it available, including mechanistic experiments, mediation and moderation analyses, regression analyses and simple correlation analyses. I conclude by discussing the problem of mechanism identity, offering a solution that relies on purpose-dependent mechanism kinds.