We study the effectiveness of a community-level information intervention aimed at improving sanitation using a cluster-randomized controlled trial (RCT) in Nigerian communities. The intervention, Community-Led Total Sanitation (CLTS), is currently part of national sanitation policy in more than 25 countries. While average impacts are exiguous almost three years after implementation at scale, the results hide important heterogeneity: the intervention has strong and lasting effects on sanitation practices in poorer communities. These are realized through increased sanitation investments. We show that community wealth, widely available in secondary data, is a key statistic for effective intervention targeting. Using data from five other similar randomized interventions in various contexts, we find that community-level wealth heterogeneity can rationalize the wide range of impact estimates in the literature. This exercise provides plausible external validity to our findings, with implications for intervention scale-up. JEL Codes: O12, I12, I15, I18.