Google Claims Click Fraud is 0.02% of Clicks

Google recently revealed how they monitor click fraud through a series of filters, offline analysis, and investigations into reported activity. In it, Google reports that they are quite proactive in stopping click fraud clicks before they’re ever charged to advertisers.

According to their model, less than 10% of all clicks are invalid, and only 0.02% get through their filters and detection system.

To understand what they’re saying, we have to make a distinction between “detected” and “undetected” click fraud. “Detected” click fraud will fall into that 10% that they catch – and don’t charge advertisers for – and that 0.02% that they miss, but is reported by advertisers – that they reimburse advertisers for.

“Undetected” click fraud are all the clicks that both Google and advertisers miss. Danny Sullivan has a great explanation of the system on Search Engine Land.

Of course, this has sparked all sorts of discussion around the net on just how accurate these numbers are. What about click fraud monitoring sites like ClickFraudIndex, which report that the industry average for click fraud is almost 14.5%? Are they just not taking into consideration the clicks Google proactively stops?

Critics are skeptical, saying that 0.02% is just smoke and mirrors – and a lot of PR. Jakob Nielsen (as quoted by ZDNet – it doesn’t seem to be on his website) put it best:

The only valid percentage is the number of manually-discovered fraudulent clicks divided by the number of clicks in those campaigns that complained and were investigated. Presumably, most campaigns don’t complain and thus are not investigated, meaning that they hide some additional clicks that would have been found through manual investigation.

The true percentage should be derived by taking a random sample of campaigns, whether or not they have complained, and investigate them manually.(Only investigating campaigns that complain might bias the sample, if you assume that some companies have the ability to estimate the extent of fradulent clicks in their campaigns.)

All of this, of course, assumes that the manual check is in fact capable of identifying the fraud.

Until that happens, we’ll continue to get conflicting data on just how much of a problem click fraud is.

Random Posts