Why We Can’t Trust Click Fraud Numbers
Tuesday, January 30th, 2007;
-- Andy Beal |
Back in December, I caused a little bit of a ruckus when I posted information from Google that suggested click fraud rates were a fraction of a percent. The majority of readers and linkers suggested that the rate was far too low and that you couldn’t trust data supplied by the search engine.
Fast forward six weeks and we find ourselves confronted with new data from Click Forensics that suggests industry click-fraud rates have increased to 14.2 in Q4, versus 13.8% in Q3. Bids for amounts over $2.00 achieved click-fraud rates of 20.9%.
So let me ask you this, who trusts these numbers? If most of you don’t trust numbers supplied by Google, can you trust numbers supplied by a company that has a vested interest in seeing click fraud grow? They may counter that they sampled 3,000 advertisers, but surely the sample set is biased. If you walk into a hospital and ask 3,000 patients, “Who here is feeling unwell?”, wouldn’t you expect a biased result?
So where does that leave us? It leaves us desperate for a third-party audit of click fraud, completed by a group that has no skin in the game. In the meantime, let’s assume click-fraud is somewhere between 0.2% and 14.2%.
Similar posts you may like...
Save & Share: 
Category: Research, SEM Industry
Forward: Email This Post
January 30th, 2007 at 11:42 am
I have just started ignoring all the noise and I pay attention to only one thing, ROI. If I had some obvious clickfraud I would take it up with Google but for the most part the campaign goes up, we track the ROI and it stays or goes based on that. Just as with retail stores we have to assume some percentage of theft into the bottom line while at the same time doing what we can (which isn’t much) to protect ourselves.
January 30th, 2007 at 6:35 pm
In general, I place my faith in security and trust analyses, which say
that as currently deployed, PPC advertising is highly vulnerable to
fraud, because of the openness of the Internet architecture, among
other things. For this reason alone, advertisers should be wary, and
consider alternatives (such as fixed fees) if they find they aren’t
getting the ROI they expect.
FYI, Vint Cerf (of Google) recently warned that 25% of all PCs may
belong to botnets. Botnets are known to generate spam, DDoS attacks,
and other things … including click fraud.
January 31st, 2007 at 4:48 pm
[...] It is understandable that the statistical analysis is quite vital for search engine advertising and Internet marketing, yet as Mr. Beal also questions the fact as to “how we can trust click fraud numbers” us another story? [...]
January 31st, 2007 at 6:42 pm
It seems to me that if you do your homework and you know for sure that:
A) You are spending $1 and you are making $10, it’s a good deal or
B) You are spending $10 and making $1 it’s not a good deal.
Either way if you know what you are spending and what you are making, it doesn’t really matter what the click fraud numbers really are.
February 1st, 2007 at 3:28 pm
I can’t come to understand how it is so difficult to know how extensive click fraud is. Jeremy Luebke and Jim Lillicotch both hinted at the answer. The advertisers are answering the question every time they bid on the PPC. If they truly felt as though they were losing money then they would stop purchasing the ads. But as proven by the fact that Google continues to make more and more money every single quarter the companies purchasing Google’s ads have no issues with the results. So Google must in turn have click fraud under control. Or there are a lot of companies out there paying Google for nothing and last time I checked people don’t usually just give away money.
February 1st, 2007 at 4:28 pm
Jason,
The answer to your question is that advertisers have not been bidding
rationally on clicks. However, they are starting to.
February 2nd, 2007 at 3:51 pm
[...] January 30, 2007: Mr. Beal had another post asking us to question that how we can trust ClickForensics results. When this company has vested interest in increased click fraud. Here is an excerpt from his post; [...]