Should We Believe Google’s Click Fraud Numbers?

Today, Google finally came out with some solid numbers around the click fraud issue. The number of invalid clicks across the Google network? Less than 10%. The total amount of undetected click fraud that advertisers have reported and asked for a refund for? .02% I was briefed by Google little while ago about their plans around click fraud and so I had some time to digest the numbers and think about them. Google also passed my name along as an expert third-party that the media could contact to get more commentary about the numbers and Google’s product roadmap for dealing with click fraud. If you’re interested in what the numbers actually mean, I would suggest going to Danny Sullivan’s post this morning on Searchengineland. Danny does his usual thorough job of making sense of the announcement.

One question that I got from a couple reporters yesterday was, did I believe Google’s numbers? Although I should have anticipated this question, I was somewhat surprised. So last night I thought about. What would Google gain by fudging the numbers at this point? I think there’s a few points you have to consider when looking for the answer to this question. Based on the fact that I’ve already been asked it three times, by three different reporters. I believe it is a valid question and one that a number of people will probably be asking.

Maybe I’m naïve, after all, I am an Alberta farm boy at heart, but in all my interactions with Google I have to say, Google just doesn’t work this way. Google is a very cautious company when it comes to divulging information. I would think one of the biggest frustrations that Shuman Ghosemajumder has had in the past is having to keep his mouth shut while various inflated numbers around click fraud were thrown about. My belief is that it’s been Shuman lobbying inside of Google that finally convinced them to open the box a little bit on the scope of click fraud in the Google advertising network. Maybe the “don’t be evil” motto of Google sounds trite to some, but people at Google believe it and take it to heart.

What would Google have to gain by releasing false numbers about click fraud? The only possible motivation would be to; one, artificially inflate their stock price, and two; encourage more advertising revenue by falsely reducing the sensitivity around the click fraud issue.

Let’s deal with the first point. I talk to financial analysts all the time and frankly, it’s been a long time since any of them asked me about click fraud. As far as a sensitive issue, there are a lot of other factors that financial analysts are looking at much closer when it comes to making recommendations on buying or selling Google stock. I believe click fraud has been already factored into the valuation and analysts have moved on.

When it comes to advertisers, there still is sensitivity around the click fraud issue, but it has lessened in the last year. The recent SEMPO study shows that as a concern for advertisers it actually trended down from 2005 to 2006. Certainly it’s something we should be aware of and keeping our eye on, but I really don’t believe it’s preventing advertising revenue from flowing into Google at this particular point. So any short-term gain that might come to Google from falsely announcing numbers could potentially be a bit of a spike in their stock price. But within a day or a week other factors would smooth that out and it would basically become a nonissue. I really don’t believe it would have any impact on advertisers at all. Short-term gain would be minimal at best.

But, the long-term cost to Google could be tremendous if they were caught releasing false numbers around click fraud. It would just be a really, really dumb thing to do, and you can say what you want about Google, but one thing they’re not is dumb. So do I believe the numbers? Yes, I have no reason not to.

The other question that the reporters asked me was what I believed the number to be for undetected click fraud. One of the reporters was actually from BusinessWeek, and if you’ve read my blog you know that I’ve taken some exception to BusinessWeek’s reporting around click fraud and search in the past. I did happen to mention that to the reporter I was talking to. The way I answered that question for BusinessWeek was that obviously we don’t know what we don’t know. Potentially there could be a lot of click fraud that slips through all of Google’s filters and slips past the advertiser as well. But for it to do so it would have to be click fraud at a extremely sophisticated level. Let me explain why it’s highly unlikely that there’s a large percentage of undetected click fraud in Google’s advertising network.

First of all Google has a number of signals they can watch to determine if click fraud is happening. Shuman mentioned that there’s well over a hundred data points they look at, including overall ROI rates, impression rates, click stream activity, click patterns, IP detection and that’s just a few of them. Also, Google is very, very good at building systems. Their engineers are the best in the world. So if they throw their collective brain power at a problem, you can be pretty confident that they’re going to come up with a robust solution to that problem. Click fraud was one of the biggest threats that Google faced in the last few years. They knew they had to restore advertiser confidence around the click fraud issue. So they threw their full engineering horsepower at the problem to build the filters that they currently have in place. This is the first line of defense against click fraud. The vast majority of the invalid click activity that’s happening in the Google network is caught by the filters. That’s the first screen that this activity would have to pass through.

The second screen is Google’s post-click review screen. This is where they look at questionable activity that made it through the proactive filters, do further investigation, and if they feel it’s warranted, they will go back and make a refund to the advertiser without the advertiser having to take any action at all. Again this is a very robust program that Google has put in place. This represents the second screen that fraudulent activity would have to get through.

The third screen in the advertiser themselves. Think about this. We have a lot of very sophisticated advertisers who have put robust analytics in place and have a deep, inherent understanding of what their website traffic patterns should look like. These advertisers have also been exposed to the so-called reporting of click fraud, like the BusinessWeek expose. They have a heightened sensitivity to click fraud so they would be very vigilant, particularly on any traffic that was coming from Google. So undetected fraudulent activity would have to get past this screen as well.

Finally, as an overall metric, Google aggregates the conversion data from advertisers who have Google analytics in place and uses that as a baseline of what typical behavior across the network should look like. In aggregate form the data allows them to check out anomalies in the dataset that may indicate fraudulent activity. This level of detection is over and above all the other fraud detection I previously mentioned. It acts as a monitor on the overall activity that could potentially indicate undetected click fraud in the network. So the likelihood of there being a significant amount of undetected click fraud is very, very low. Once again, so low it’s probably not worth spending much time worrying about.

The gist of my column today in SearchInsider is advising advertisers to look at a much bigger picture than just focusing on click fraud. I realized Google had to release these numbers because everyone was asking for them. If we can accept those numbers than perhaps we can get on with looking at our overall campaign performance and really spending some time on the things that would have a much greater impact on our overall return on investment. For example, the drum that I will continue to beat as long as anyone is willing to listen is for advertisers to focus on their own conversion rates. Time after time, I see landing pages that aren’t optimized and aren’t aligned to the intent of the potential visitor. I see sloppiness in advertising messages with a lack of relevancy aligned to the queries that are used. If advertisers paid more attention to these things they’d realize far greater benefit than they would by fretting over click fraud.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.