Google Data

Posted by Hal Varian, Chief Economist

There has been some recent discussion in the press about the potential impact of our advertising agreement with Yahoo! on ad pricing, and we wanted to help clear up a few misconceptions.

Some of those misconceptions appear to be based, in part, on a July report by SearchIgnite that concluded, among other findings, that once the agreement is implemented, keyword prices on Yahoo! might increase by an average of 22%.

After taking a close look at the study, I believe it makes several flawed assumptions and uses questionable methodology. The paper suggests that advertisers will be getting the same performance from the same ads, just at higher prices. We believe that advertisers will be getting significantly better performance at prices that reflect that improved performance.

Let's take a look at some problems with the SearchIgnite report.

First and most importantly, the report fails to acknowledge that ad prices are not set by Yahoo! or Google, but by advertisers themselves, through the auction process. Since advertisers set prices themselves via an auction, the prices must ultimately reflect advertiser values. That process will remain completely unchanged by our agreement.

Second, the report mistakenly claims that for any given keyword, Yahoo! will have the ability to see whose ads are priced higher -- Yahoo's or Google's -- and then decide which ads to serve. In fact, under our agreement Yahoo! won't be able to see the current auction prices for Google ads, just as Google won't be able to see Yahoo's prices.

Third, the report mistakenly assumes that Yahoo! will serve Google ads for as many of its search queries as possible. This contradicts Yahoo's own statements that their plan is to serve Google ads on search results pages where they have few relevant ads to serve. Yahoo! also has a strong economic incentive to keep serving as many of their own ads as possible, since they get to keep all of the revenue from those ads, while Yahoo! only receives a part of the revenue from ads served by Google.

Fourth, the report includes a misplaced focus on cost per click (CPCs) rather than the more important measure for advertisers -- return on investment of their advertising dollar. One of the reasons Google's ad system has performed so well for advertisers is that our ads tend to be highly relevant to user queries, which makes it more likely that a user will click on an ad and purchase the advertiser's product. We have found that advertisers are generally willing to pay more per click so long as those clicks result in more sales. We anticipate that our agreement with Yahoo! will bring more relevant ads to Yahoo! users -- which is better for both advertisers and users.

Finally, the report suffers from a number of methodology flaws. For one, the study fails to take into account that fact that Yahoo! shows significantly more ads per page than Google. Since both search engines tend to show higher cost-per-click ads in higher positions, showing more ads automatically tends to reduce the average cost-per-click. Also, the study's terms are vaguely defined. Its authors discuss "head" and "tail" keywords, for example, but never clearly define what they mean. Are those terms that appear less often than once a day? Or once a week? There's a big difference.

As we have said before, Google doesn't set advertising prices -- advertisers do. Prices must reflect how much a sale is worth to an advertiser, and that will continue to be the case after our agreement with Yahoo! is implemented.