We examine how allocating a fixed social assistance budget across communities using different poverty indicators affects both beneficiary selection and program effectiveness. Our study focuses on a year-long unconditional cash transfer program in Lebanon that implemented a nationwide randomized policy experiment to test the implications of using four alternative geographic prioritization rules that target monetary poverty, food insecurity, nutritional deficiency, and multidimensional deprivation. By simulating each allocation rule for every household, we directly identify marginal beneficiaries-approximately 35% of the population in our context-whose assistance varies across targeting rules. There are significant differences in poverty rates, demographics, and market access among these households. We then recover local average treatment effects to show that the program alleviates economic deprivation, with effect sizes ranging from 0.02 to 0.21 standard deviations depending on the outcome and marginal beneficiary population. Children’s school enrollment increases by 7–9 percentage points exclusively among beneficiaries marginal to consumption-based poverty targets. Program satisfaction and perceptions of fairness are determined by the level of benefit received, but not by the targeting strategy used. Focus group discussions and survey data suggest that heterogeneity in outcomes arises from aggregate shocks interacting with existing market failures faced by households. District-level analysis shows substantial variation in program effects, but limited ability to target effectiveness using data typically available to implementers.