Electrical utilities offer incentives to their customers to reduce their demand during temporary supply-demand mismatches. While customers would prefer a higher incentive to participate, utilities would prefer to minimize the incentive while achieving a target reduction. Because the incentive affects the bottomline of the utility, identifying the optimal incentive reflecting this trade-off is important. Several works have focused on how to implement DR in a building, but there has been little work on identifying the optimal incentive from the utility’s perspective. We complement existing work with an approach on how a utility can identify the optimal incentive for a set of buildings that it serves, while meeting individual buildings’ constraints. To this end, we build Demand-Response Potential (DRP) models that give the economically rational demand reduction of a building as a function of the utility’s offered incentive. For handling scalability at the utility level, we approximate the DRP using regression based approach. We evaluate our approach on the PLUTO dataset of building types and sizes. We find that the DRP varies with building types (from 19% for restaurants to 54% for warehouses). It is typically low for buildings with high thermal inertia; and building types with low individual DRP can contribute significantly in aggregate due to their numbers. Offering non-uniform incentives to different buildings can improve the utility’s DR benefit by up to 19% compared to offering uniform incentives.