Cost-per-Impression Pricing and Campaign Delivery for Online Display Advertising

Pricing and capacity management represent significant challenges for web publishers that generate revenues by selling advertising space on their websites. Advertisers approach a publisher to book an advertising campaign, requesting a number of impressions to be delivered regularly throughout the campaign duration. Publishers offer multiple advertising plans. They face uncertainty in demand and supply, which generates non-uniformity in the campaign delivery. Based on a stylized model of the publisher’s operation, we suggest a capacity allocation mechanism parameterized by a display frequency that allocates viewers to ads in a rotating manner. Through a large-capacity system analysis and under the suggested mechanism, we prove that the fluid price and display frequency are asymptotically optimal. We also obtain correction terms for the fluid solution when used under a regular regime. The pricing and display frequency can be translated into inputs to delivery engines used in practice. We obtain data from a publisher and perform an extensive numerical analysis that reveals the interrelation between prices, traffic load, impressions, and display frequency.