Abstract Marginal pricing has long been the instrument of choice to address water conservation challenges. More recently, non-price behavioral interventions have emerged as an alternative. However, there is limited data on the relative efficacies of price and non-price interventions. We report results from long-term field experiments studying unit-level water conservation responses to both price and non-price interventions in the same group of households (n = 64,186 household-days). Conservation habits, attitude-action gaps, principal-agent incongruities, and billing cycles help account for the heterogeneity in response between households, and across time. A non-price behavior modification intervention before the introduction of marginal pricing resulted in a large and significant effect on treated households (33%). The subsequent introduction of marginal volumetric pricing also reduced water use (8%, for previously untreated households). However, this average price effect masks how a large share (21%) of households increased water use, or how a mere 12% of the households accounted for all the aggregate reduction in water use. We investigated such heterogeneous responses as a systematic conservation maximization design question beyond statistical variance in individual responses. We used daily water consumption measurements across three years alongside a household survey to delineate structural and agentic barriers to conservation behavior. Our analysis reveals how combining price and non-price behavioral interventions could hold the key to achieving conservation effects that are both large and persistent.