Optimizing Reserve Prices in Display Advertising Auctions

This paper considers how a publisher should set reserve prices for real-time bidding (RTB) auctions when selling display advertising impressions through ad exchanges, a $115 billion market and growing. Conducting a series of field experiments to induce exogenous variation in reserve prices at a major publisher, we find an increase in the publisher's revenues by 35% and observe evidence that advertisers use a minimum impression constraint to ensure advertising reach.Based on this insight, we construct a structural model of advertiser bidding model that accommodates impression constraints to infer the overall demand for advertising as a function of reserve prices. Using this demand model, we solve the publisher pricing problem. Consideration of minimum impression constraints in setting reserve prices generates a predicted increase in profits of nine percentage points over a solution that does not incorporate the constraint. In a final series of field experiments, we validate our model's predictions by showing ad revenues tend to be highest across exogenously varied reserve price levels closest to the imputed optimal reserve prices