• ZDNet, Monday, March 5, 2007 10:45 AM

Google consultant Shuman Ghosemajumder published data back in December stating that the rate of click fraud on Google was just 2 percent. This purred the Web giant to internally investigate the matter. The result is some new figures and new tools the Web giant rolled out last week to fight click fraud. For those who missed it, advertisers can now block ads from being shown to certain IP addresses. Additionally, Ghosemajumder found that the average rate of invalid clicks already filtered by Google is less than 10 percent, while the rate of those that slip through the system's cracks a miniscule .002 percent -- 1000 times less than in December.

ZDNet writer Donna Bogatin takes serious issue with the methodology Google uses to get that figure. "Google is NOT affirming that for every ten thousand clicks on AdWords ads, fewer than two are fraudulent clicks," she says. Rather, for every ten thousand AdWords ads, the number of user-submitted clicks Google AGREES represent click fraud amounts to two.

Therefore, it's misleading to conclude a .002 percent fraud rate, because there's no accounting for the number of clients that haven't investigated the issue, in addition to ignoring claims where Google decided against reimbursement. In the interest of neutrality, a more accurate figure would be to take a random sample of advertisers and manually investigate the issue.

Read the whole story at ZDNet »