Buy-it-now or Take-a-chance: A New Pricing Mechanism for Online Advertising

Abstract

Increasingly sophisticated tracking technology oers publishers the ability to oer targeted advertisements to advertisers. Such targeting enhances advertising eciency by improving the match quality between advertisers and users, but also thins the market of interested advertisers. Using bidding data from Microsoft's Ad Exchange (AdECN) platform, we show that there is often a substantial gap between the highest and second highest willingness to pay. This motivates our new BIN-TAC mechanism, which is eective in extracting revenue when such a gap exists. Bidders can \buy- it-now", or alternatively \take-a-chance" in an auction, where the top d > 1 bidders are equally likely to win. The randomized take-a-chance allocation incentivizes high valuation bidders to buy-it-now. We show that for a large class of distributions, this mechanism achieves similar allocations and revenues as Myerson's optimal mechanism, and outperforms the second-price auction with reserve. For the AdECN data, we use structural methods to estimate counterfactual revenues, and nd that our BIN-TAC mechanism improves revenue by 11% relative to an optimal second-price auction.