Click fraud is in the news again this last week, thanks to a couple of stories. The first was that a company called Outsell claimed that
click fraud cost advertisers $800 million last year. The other was that Eric Schmidt evidently
dismissed the issue at a conference last month.
The issue once again returns: is click fraud a big deal or not. My conculsion has always been that click fraud is bad, primarily becuase it undermines the confidence that advertisers have in search advertising. However, it seems to be a very solvable problem, at leat on an advertiser by advertiser basis.
Let's take that first story. That number is large and impressive, but really pure speculation. Educated speculation, clearly, but there's no column on the ledgers marked "Click Fraud (loss)." Rather, this is the result of asking advertisers to guess at the amount of clicks they get that are fradulent. The problem with this is that the majority of advertisers have no sophisticated means by which they can measure their traffic, making it difficult to pull out the percentage that may be fradulent.
So, that $800 million number is really more a measure of how concerned people are about click fraud, which is a TON.
Making Dr. Schmidt's remarks a bit ill-advised. The feedback on his comments tend to be just that: he seems arrogant and unconcerned. Well, chalk that up as one more Google-exec remark-snafu.
But, on the other hand, I think he's kidding himself when he says that the perfect economic solution is to ignore it, and let the market take care of things. Take a look at any market: all attempts at deregulation, under the assumption that a free market naturally tends toward efficiency, seem to go horribly wrong. Sooner or later, a number of rules need to be put in place that force the market back to 'correct'. Free market solutions seem to never account for the presence of black markets.
The situation, then, is that advertisers are worried, but unclear on the details (which, of course, fuels the worry). In this case, I agree with Aaron from
SEOBook:
If you believe the advertisers will filter out the fraud then at least bake the ability to detect value on a per publisher level.
Maybe what we need is a greater level of detail for everyone. Google's awesome, but its inner workings are obscure. That's acceptable to a level. I think the algorithm should be secret. But the flow of clicks for a publisher? That is valuable information and should be exposed to a greater degree to the publisher. That will allay some fear, which will go a long way toward solving the issue.