Google recently posted on their Adwords blog about their overall numbers for invalid clicks and posted about a few upcoming features as well.
After reading this I can say that the .02% click fraud rate may not usually be what you can expect when looking at your particular vertical. Below I will cover a few topics of interest from their post.
The Invalid Clicks Number Provided in AdWords Reporting
The Invalid clicks report – is a number (with no supportive data provided) of the amount of invalid clicks (click fraud included) that Google says they have filtered out available in AdWords reporting.
Even if this number were accurate it would only represent the amount of invalid clicks that Google is able to detect. If you have done any reading on the technology in use to wage fraud then you know it doesn’t look good. I wont divulge these practices here in the hopes of not further spreading them.
One of my favorite statements from the blog post is (and that Yahoo threw out in a SES click fraud panel I was on):
“Advertisers have always been able to compare their log data with their AdWords charges to calculate an estimate of the number of invalid clicks in their own account.”
1) No two analytics programs ever match even when having the same source data.
2) If they did match, the best I would know is how much more or less I am off from the total amount of clicks reported by AdWords. The AdWords system, like every system is not perfect and doesnâ€™t always count clicks properly. In addition, once an ad is clicked there are reasons why that click would never appear in my server logs – a few examples off the top of my head are:
- The user may click back on their browser before they get to my site
- There was a network error along the way to the site and the user never got there
- The http request header has been altered by the user (more and more users are wanting more privacy)
- There was an internal issue with AdWords or their partners
On the Future AdWords IP Blocking Feature
I think this is more of a feel good move for advertisers than anything. Many advertisers initially think the biggest threat to their PPC campaigns is a direct attack on their PPC budget through clicks on their ads directly from advertisers in their offices. This is most often not the case and even so is the most basic form of click fraud and the most detectable, which is easily filtered. As an advertiser it may not be wise to start shutting down proxy IPs of AOL, Earthlink, Verizon, etc.
Activity Level Numbers
“The most significant similarity is that the seriousness of the problem is not measured by how much spam is sent, but rather how much gets into a userâ€™s inbox.”
Google tries to compare click fraud to email spam saying that spam only matters when it hits your inbox and that’s the benchmark we should be using. The difference is that is that for spam I am on the receiving end of email spam, not the advertiser with marketing capital at stake.
“When looking at click fraud, the most important measure is not the “activity” metric which “measures the volume of invalid clicks that occur overall” but the “impact” metric.”
That’s like saying in the old west it shouldn’t matter the amount of thugs and robbers and murders there were looking to rob stagecoaches, it only matters that you have someone riding shotgun. Of course stagecoaches with guys riding shotgun were seldom-robbed….right.
When it comes to activity Google just says it averages out to 10%. 10% of what number? How does that number reflect on particular verticals? It makes sense that some verticals may see little activity and other verticals where the cost per click gets a little higher and where the traffic is stronger could be hammered, respectively.
“Because it is difficult to definitively determine the “intent” of a click in many cases, the number of invalid clicks that we filter also include those filtered for reasons separate from fraudulent intent.”
If that’s true then I guess the reverse is also true: they have difficulty in definitively determining if a click is fraudulent.
“For example, we have an automated rule which filters out the second click of all double clicks as a matter of policy. We mark this kind of activity as invalid simply to optimize advertiser ROI.”
That wasn’t always their policy. It didn’t change to about two years into the AdWords program (http://googleblog.blogspot.com/pdf/Tuzhilin_Report.pdf).
The Meat: The .02% Click Fraud Rate
What is this number really? Google says this number is the result of:
- An advertiser submitting a request for review of suspicious activity and provides a data sample
- Google reviewing that data
- Google siding with the advertiser
A couple of things to ask yourself:
1) How many advertisers are actively monitoring their campaigns for click fraud?
2) How many times do advertisers agree with Google’s final determination?
If you’ve read what has been published by SEMPO and been to a few SES panels on this, you would know that many advertisers do not actively monitor and audit their PPC campaigns and that there is usually a strong discord with those who have tried to deal with Google in the past on suspicious activity with their accounts.