Most electoral fraud is not conducted centrally by incumbents but rather locally by a multitude of political operatives. How does an incumbent ensure that his agents deliver fraud when needed and as much as is needed? We address this and related puzzles in the political organization of electoral fraud by studying the perverse consequences of incentive conflicts between incumbents and their local agents. These incentive conflicts result in a herd dynamic among the agents that tends to either oversupply or undersupply fraud, rarely delivering the amount of fraud that would be optimal from the incumbent’s point of view. Our analysis of the political organization of electoral fraud explains why even popular incumbents often preside over seemingly unnecessary fraud, why fraud sometimes fails to deliver victories, and it predicts that the extent of fraud should be increasing in both the incumbent’s genuine support and reported results across precincts. A statistical analysis of anomalies in precinct-level results from the 2011–2012 Russian legislative and presidential elections provides preliminary support for our key claims.