Last week, Wendy Davis in Just An Online Minute reported on Google's analysis of click fraud -- or, more so, its assessment of the lack thereof. After users lodged complaints with Google regarding click fraud, its "reactive" invalidation of clicks was approximately 0.02%, according to a post on the Inside Adwords blog that Davis cited. With such a minimal refund, it is clear why advertisers would be frustrated with Google and its search engine brethren when it comes to this touchy subject.
While click fraud continues to get tremendous media attention, impression fraud, the more dangerous of the two, is pushed aside. As defined at Wikipedia, "Impression fraud is an insidious variant of click fraud in which the advertiser is penalized for having an unacceptably low click-through rate for a given keyword. This involves making numerous searches for a keyword but without clicking of the ad." While this is a good starting point for discussion, impression fraud takes many forms.
As most advertisers now know, the Google Adwords algorithm had a "technical issue" on Feb.16, as announced on Inside Adwords on Feb. 20. The same day of the "technical issue," our clients saw an increased impression level of 30% over expected levels, along with a 22% decline in click rates. These dramatic numbers force advertisers to ponder the realities of the "technical issues" and whether it was due to the noted "routine infrastructure update" -- or something deeper.
Over the past few months, we've seen impression fraud take new form with great impact. Recently, a client saw the following performance occur on Google with its trademarked brand terms:This amounted to a whopping 1300% increase in impressions with correlative CTR drop of 95%, causing a 775% increase in CPC charges and 932% increase in Cost to Acquire. Our internal analysis showed that the fraud was triggered by an affiliate that targeted "broad match" iterations of branded terms while the affiliate conducted an "exact match" on the same terms. Google disagreed, and unfortunately, because we cannot gain access to the impression logs, we have to take them at their word that the price increase was justified and that this was a matter of poor marketing, not a Google system issue.
Cybersquatting (a form of impression fraud) now appears to be seeping into the Google Adsense network as well. With Google Adsense's site targeting features, advertisers can now see that some of the better performing Web sites in the network happen to be domain parking Web sites. Most particularly, some of the best performers appear to be cybersquatters with misspelled domain focuses. Given advertisers' disdain for cybersquatting and domain parking, has anyone told Google that these may be undesirable strategic partnerships? While Adsense performance has been increasingly improving the past few months, what is the cost to the advertiser?
Finally, you cannot discuss impression fraud without bringing up search arbitrage. While here is some appropriate effort to stop search arbitrage, the majority is inadequate. Particularly, the most-often-seen deception is a paid search ad that miscommunicates the offerings actually found on the landing page. This, by definition, is a form of fraudulent impression.
So, why is this a big issue? Mainly because Google and others who share their ads have strict language in their agreements with ad distributors that prohibits fraudulent impressions. The reason for this firm language is due to the impact that fake impressions can have on CPC rates and "invisible environment" quality scores. In summary, fraudulent impressions equal lower quality scores, and thus, increased CPC rates.
It's time for fewer public relations efforts and more self-regulation and policing by the search engines, to truly minimize the threat. Advertisers are more interested in this than refunds.