The Click Fraud Conundrum

To determine if a click is fraudulent or invalid, one needs to define a click. Easy, right? Perhaps not. The reports on the magnitude of click fraud vary dramatically. Similarly, the data on overall click quality and which factors tend to correlate highly with quality are scarce (many organizations, including my own, have done their own analyses). Everyone's jumping on this bandwagon. So today, I'll focus on the most basic premise: the click. What is a click and when should a marketer be charged for one? Similarly, when should publishers be paid for one?

The crux of the click fraud issue is marketers shouldn't have to pay for clicks that aren't authentic or valid. A tangential issue is what other clicks may not be suitable as media and, therefore, shouldn't be paid for. The goal seems to be to define and determine what a billable click is. Even if we can achieve a black-and-white definition of a click or a billable click, we're left with the challenge of measuring and validating those clicks.

Toward that end, several major Internet advertising players are attacking these issues. The Search Engine Marketing Professional Organization (SEMPO) is supporting a Fair Isaac Corp. study of click quality. It will attempt to understand the click fraud and click quality problem. Researchers are calling for participants, so if you want a personalized, confidential report, whether you're the advertiser, a Web analytics provider, or an agency, follow the link above and consider participating. You'll also gain access to the public report.

The Interactive Advertising Bureau (IAB), which championed and authored the industry standards for counting ad impressions, also launched an initiative to "create a set of Click Measurement Guidelines." Clearly, loss of marketer and advertiser confidence in the validity of clicks and concerns about the scope of the click fraud problem were major catalysts in moving this project forward. The reality is the IAB committees, including the Search Engine Committee (of which I'm a member), have been discussing the need for clarity in click auditing, reconciliation, measurement, and definitions for some time.

No doubt Google, Yahoo, MSN, Ask, and the other search engines have learned more about both advertiser perception of the click fraud issue and the confusion in the marketplace brought about by lawsuits and an explosion of companies offering pure click fraud monitoring solutions. All the major engines are big supporters of the IAB Click Measurement Working Group, as are the third-party ad-serving, click-counting, and campaign management companies. Impression fraud and the fuzzy way impressions were sometimes counted nearly derailed the online advertising business in its first decade. Similarly, click quality and click measurement standards are required to ensure all parties engaged in SEM (define) are on the same page (pun intended).

Google even went so far as to start showing advertisers the valid and invalid clicks in AdWords reporting. It released a white paper (more of a research report) criticizing many players in the click fraud monitoring business and click fraud consultants for making fundamental analysis errors resulting in an overstatement of the click fraud problem. The report even named companies whose estimates Google claims are flawed.

Interestingly, some of the report's points will be the same issues Fair Isaac will have to reconcile when reviewing participant data. The biggest issues are undifferentiated visits and use of log files as a data source when counting inbound clicks. We have found inbound ad click redirects are the most accurate method for counting clicks and know their source and frequency. Clearly, Google isn't planning to issue credits to advertisers based on analysis it considers flawed.

In conclusion, perhaps I can start a dialogue on what exactly a billable click is. I don't expect this column to serve as an agenda point for any of the initiatives already underway or new ones that may be beginning. This is more stream of consciousness (define). Let's start by listing the elements all parties can likely agree define an advertising click:

• A click is generated by a human person, not a robot or automated spider.
  
  
• When a human creates a click, she does so consciously and with purpose.
  
  
• Humans who click are interested in navigating to a specific page or site.
  
  
• Humans understand clicking is a navigational action resulting in the browser loading the new page or site.

More nebulous areas relate to when a human-generated click with all the above intentions shouldn't be billable. Should a click be billable if:

• The person repeats the navigational behavior on the same exact page (SERP), resulting in a second click within a session or short period?
  
  
• The person clicks a link to the same advertiser on a different SERP or from a different contextual or behavioral ad impression?
  
  
• The person is the advertiser's competitor and doesn't have a genuine interest in the ad?

If any of the above or other conditions invalidate a click, does the publisher, network, third party, or advertiser have the responsibility to reconcile the billable and invalid clicks? My guess is much of the work done by the IAB in defining and standardizing impressions will be quite valuable in guiding the next round of standards-setting.

Stay tuned. This discussion is far from over.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.