We consider the potential of randomized encouragement designs in agricultural development. The classical problem in program evaluation is the inability to observe individuals both participating and not participating in a program at any point in time. Randomized encouragement designs solve the evaluation problem by randomly varying incentives to participate in a program without affecting outcomes of interest, thus making it possible to measure average treatment effects. They are also of special interest to agricultural development economists for their potential to stimulate program participation; low participation is common in many programs, making it difficult to estimate impacts. These benefits come with caveats that should be taken into account both in research design, and in econometric practice. We explore these issues in a context with which are familiar: an evaluation of an index insurance product utilizing a randomized encouragement design, mirroring our current work in coastal Peru.