A best-price policy (BP) is a promise by a seller to give her customer a refund if she reduces her price after the customer has already purchased the product. We characterize the optimal BP policy as when the seller can control both the policy length (when the promise expires) and the refund scale (what portion of the price difference is refunded). We explain why the policy length is finite and varies across industries. In a setting where consumers' valuations decline over time, we show that a finite-length BP allows the seller to commit to not lowering her price too soon, while at the same time letting her capture some of the benefits of intertemporal price discrimination. However, because the decline in consumers' valuations is uncertain, a BP does not allow the monopolist to achieve the profit she could earn with a full commitment.
