In August 2007, the First International Workshop on Data Mining and Audience Intelligence for Advertising (ADKDD’07) was held so that participants could share their thoughts on how data mining and advertising interact, and address issues in this field.
One of the papers discussed during the workshop from Yahoo Research, was Pay-per-action model for online advertising, by Mohammad Mahdian and Kerem Tomak
The online advertising industry is currently based on two dominant business models: the pay-per-impression model and the pay-per-click model. With the growth of sponsored search during the last few years, there has been a move toward the pay-per-click model as it decreases the risk to small advertisers.
An alternative model, discussed but not widely used in the advertising industry, is pay-per-conversion, or more generally, pay-per-action. In this paper, we discuss various challenges involved in designing mechanisms for the pay-per-action model, and approaches to tackle some of them.
Pay-per-action (ppa) goes a step beyond pay-per-click (ppc) advertising by tying the payment of an advertiser not to the click of a link but rather an action or step past that to be taken by a consumer. This could be an actual sale, or a phone call, or even a direct online order.
The paper provides an overview of how PPA works, what advantages it provides as well as what issues arise under the use of a ppa model. It also discusses how a search engine might address some of those issues.