Profiting from Click Fraud


A day or so ago, Shuman Ghosemajumder from Google posted two blogs on his personal site questioning Click Forensics data, methodologies and motivation. While it is really unfortunate that we have to take time to address these issues, I will. It is unfortunate because Shuman and I have had numerous phone conversations, email exchanges and meetings over the past nine months and we discussed many of the topics he brought up in his post.  Let me begin…

1) Many third parties have not counted clicks properly Click Forensics counts clicks properly. We do not count back clicks. We count page views correctly. As a matter of fact, we use click ID’s (including Google’s) and other methods to ensure that our reports are the most accurate in the industry. 2) Inflated click counts result in even more inflated “click fraud” estimates Google calls them “invalid clicks.” Click fraud is a subsection of this category and does include malicious attacks, competitive clicks and affiliate fraud. Shuman has said on many occasions that Google does not charge for invalid clicks. This is a game of semantics. Our report focuses on what we call “high threat level clicks.” These are clicks that our rating engine and algorithm identified as being unwanted by the advertiser. These are clicks that advertisers should not be paying for and the reality is that in Q4 of 2006 14.2% of all clicks we scored fell into this category. 3) Even if they fixed those problems, they’re not actually measuring click fraud Actually the truth is Click Forensics measures and reports on all suspicious click activity and attempted click fraud across search engines and their content network advertising channels. Google has admitted that is has not charged for some clicks. Shuman told Andy Beal that this number was .2% of all clicks.  Our advertisers (and Google’s) can see the number of clicks that they credited and it is far lower than the number of unwanted clicks search engines credit them for in their campaigns. As a matter of fact, many of the 3,000 advertiser members of the Click Fraud Network report that they have never been discounted more than 2 percent for the invalid clicks. It is clear search engines are not catching all bad click activity. Part of the reason is their lack of advertiser-side information and the difficulty in policing content networks alone. 4) Industry metrics are not necessarily the same as Google’s metrics True.  Our industry metrics a representative of virtually every search provider large and small in the industry. This is why we have a statistically accurate view of the true rate of industry click fraud levels. We used to separate our click fraud figures by search engines tiers but stopped reporting this way this quarter because there is a growing disparity between individual search providers as to the click fraud levels. We didn’t think it was fair to lump the good with the bad so we will begin to report on individual providers later this year. The results may surprise you. 5) ROI on the content network is the same as it is on search That is an interesting way to spin the story.  The fact is that content networks produce a higher amount of fraudulent and invalid click activity.  We reported that 19.2% of clicks coming from content networks are fraudulent and fall into the high threat category. I’m glad Google agrees with us on this one.  Its clear advertisers have been justified in their views on the click quality of content networks because the threat level is so high. We help advertisers identify publishers in the content networks that they should avoid. This approach actually helps improve the ROI on all content networks and gives advertisers peace of mind, thereby accelerating the growth of pay per click advertising. Shuman’s first post ends with this, “The key point here is not that their (Click Forensics) numbers are "too high". The point is that their data collection methods are inherently flawed and any resemblance their numbers could have to reality would be coincidental. Even so, given that they are not measuring click fraud, they apparently don't intend their numbers to reflect reality.”

At Click Forensics we stand by our numbers, our methodology, our people, our clients, our partners and the 3,000 plus members of the Click Fraud Network.  And we are not alone in defending our approach.   In March 2006, our methodology was examined by Dr. Tuzhilin, the same expert who examined Google's click fraud practices and wrote the report as part of the $90 million click fraud settlement case.  Dr. Tuzhilin said:

Click Forensics has good data and this is a source of their advantage over the search engines. My role is to work with them to refine the scoring methodology to improve accuracy. Their approach is to incorporate as much data as possible to improve accuracy. The search providers simply don't have enough data to have the most accurate approach."

The final false accusation from Shuman, responding to a post, charged us with releasing data to  …promote flawed industry estimates in order to attract business.” 

Nothing could be further from the truth. 

We are not compensated based on how much fraud we find.  But each time a fraudulent click occurs and is missed by a search engine, they get paid for it.

Our goal is to continue to build a great company than helps advertisers ensure they get what they pay for in the tradition of Nielsen and Arbitron.  Google would do well to take a less confrontational and more cooperative approach embraced by their competitors.  That said, I look forward to discussing these issues with Shuman and continuing to work together to accelerate the growth of online advertising by fostering trust and accountability.