Revenue Management for a Web Publisher Using Advertising Networks

The Internet has been a fast growing advertising medium. We consider a web publisher that generates revenues from displaying advertisements on its website with advertisers being supplied through an advertising network. The advertisers request their ad to be displayed to a certain number of visitors to the website. We focus on the main operational challenge of matching uncertain demand from advertisers requesting advertising space to uncertain supply from viewers. The publisher faces challenging decisions of determining the price per impression, number of advertising slots, number of advertisements that share each advertising slot and others. Our stylized model is a novel queueing system with no waiting space (loss system), where advertising slots correspond to servers. What sets this novel system apart from known multi-server queueing systems is its service mechanism; the advertising slots act as synchronized servers. We derive a closed-form solution for the system's steady-state probabilities and determine the optimal price. We analyze the publisher's optimal decisions and for example show that the optimal price increases in the number of impressions made of each ad, which goes against the quantity-discount commonly offered in practice. We illustrate that our model serves as an effective building block for more complicated settings such as competition, for which we provide some initial results.