In Pay-Per-Percentage vs. PPC, Shimon Sandler points out an interesting new paper from the folks at Microsoft Research – Pay-Per-Percentage of Impressions: An Advertising Method that is Highly Robust to Fraud (pdf)
As Shimon notes, the idea is that this type of advertising approach would be “immune to both click fraud and impression fraud,” and would use something called “pre-fix match” instead of broad match.
The author of the paper is Joshua Goodman, who is a Principal Researcher, and the head of the Microsoft Learning for Messaging and Adversarial Problems (LMAP) team, and who has an impressive page of other publications listed on the Microsoft domain, including a recent one on Finding Advertising Keywords on Web Pages (pdf).
What does pay-per-impressions mean? Simply, someone can can for a percentage of all impressions for certain keywords or keyword phrases over a period of time.
In this system, an advertiser picks a keyword, e.g. ?cameras? and purchases, perhaps through bidding, a certain percentage of all impressions for that keyword. For instance, an advertiser might pay $1.00 to MSN Search. In return, the advertiser might receive 10% of all impressions for ?camera? for 1 week. What does this mean? It means that for 1 week, one out of ten times that someone searches for the word ?camera?, they will see the ad.
The number of real impressions that an advertiser receives would not be affected by the number of fake impressions. The paper describes how this mechanism would need to work to avoid impression fraud, and how a broad match-type of system could function under a pay-per-percentage type system.
The paper points out that if an pay-per-percentage system is adopted, that it wouldn’t replace pay-per-click or pay-per-impression, but would rather be another choice that an advertiser can make. A method is described in the paper that would allow these types of systems to function side-by-side.
It also looks at something that the author calls “misinformation fraud” which is when a competitor performs searches to boost the apparent rate of search volume that an advertising system might display as indicative of how popular a word or phrase is, and affiliate fraud.
It’s an interesting and thoughful paper, well worth a read if you use paid search to attract visitors to you web site. Thanks for pointing it out, Shimon.
Postscript: Donna Bogatin, from Digital Micro-Market also pointed this paper out on July 1st, and has an interesting post from yesterday asking Google to show that they are taking a stand against click-fraud in Click fraud prevention: The next great search engine differentiator?.
The Microsoft paper was published as part of a workshop last year sponsored by Yahoo, but really doesn’t seem to have received much attention at the time, though Dr. Garcia (Orion) wrote about it at the Search Engine Watch Forums, along with other papers from the same workshop in December, in a post titled Searcher Perceptions & Paid Links.