June 22, 2006, 4:09 PM — Google Inc. is testing an ad model that in theory isn't vulnerable to click fraud, a serious problem that some believe puts in danger Google's main source of revenue: search engine-based advertising.
Google has built its business on displaying text ads based on a keyword or phrase entered in its search engine or included on a Web page published by Google or by a partner publisher. In this model, known as pay per click, Google collects the entire click fee for ads displayed on its pages. If the ad is displayed on a third-party Web site, Google splits the fee with the publisher of that Web site.
While very successful, pay-per-click ads are vulnerable to click fraud, which occurs when someone clicks on a pay-per-click ad maliciously, without any intent to do business with the advertiser. Click fraud perpetrators engage in this practice for two main reasons. First, a company may want to run up a competitor's advertising costs. Second, a Web site publisher may want to increase its commission revenue.
Click fraud is hard to track and detect, and estimates of its incidence vary widely, with some suggesting as many as 20 percent of clicks may be fraudulent, in which case Google's business could be in jeopardy. Google derives virtually all of its revenue from pay-per-click ads. To a lesser extent, Yahoo Inc., AOL LLC and Microsoft Corp.'s MSN division also stand to lose from click fraud.
Now, Google has begun inviting some of the publishers on its AdSense Web site advertising network to test cost-per-action (CPA) ads, which are similar to pay-per-click ads but with a major difference: advertisers only pay when the click yields a specific result, such as a concrete business lead or an actual sale. Some niche players already provide CPA ads, but Google's entry into the market would bring the concept to a critical mass of advertisers.
A Google spokesman confirmed via e-mail that the company is conducting this test. "We
