Index insurance is a promising tool to reduce the risk faced by farmers, but high basis risk, which arises from imperfect correlation between the index and individual farm yields, has limited its adoption to date. Improving adoption will require reducing one or both of the two fundamental sources of basis risk: the intrinsic heterogeneity within an insurance zone (zonal risk), and the lack of predictive accuracy of the index (design risk). Previous work has focused mostly on design risk, conflating the quality of the index with the quality of the zone. Consequently, there is currently no way to distinguish a “good index in a bad zone” from a “bad index in a good zone”. Here we investigate the relative roles of zonal and design risk, with two main contributions. First, using a formal decomposition of basis risk, we show that the optimal index is the first principal component of the correlation matrix of yields between fields. This provides a simple upper bound on the insurable basis risk that any index can reach within a given zone. Second, we use 10 m resolution satellite data on maize yields in Kenya to provide the first large‐scale empirical analysis of the extent of zonal versus design risk. Our results show a strong local heterogeneity in yields, underscoring the challenge of implementing index insurance in smallholder systems and the potential benefits of low‐cost yield measurement approaches that can enable more local definitions of insurance zones.