Despite theory-based expectations, and actual usage in practice, we lack any evidence about sales productivity gains from activity-based incentive (ABI) pay. Such plans incorporate activity scores from salespersons’ call reports into incentive compensation. Securing the cooperation of a pharmaceutical firm, we undertook a three-year experiment involving an “ABI treatment-removal” design. Our first intervention added ABI pay for front-line salespeople and their supervisors across 305 sales territories. Our second intervention removed ABI pay from salespeople, and the third intervention removed them from the supervisors as well, thus returning to the status quo ante. We find a robust sales gain around 8% from each ABI intervention relative to the no-ABI baseline. These ABI sales effects are moderated by the number of salespeople in a territory with larger gains recorded in territories with more salespeople. Surprisingly, these sales gains are statistically indistinguishable regardless of whether ABI pay is provided to both supervisors and salespeople or just to the supervisors. The ABI effects on the behavioral activity measures show that supervisors exert behavior control downwards on salespeople when they (supervisors) are paid ABIs. All these results are robust to various threats to validity, including alternate specifications, attrition and accounting for the nesting of geographically adjacent territories. Managerially, our work supports tying incentive compensation to call reports despite the potential for self-serving biases in these measures, because supervisors exercise behavior control effectively.