60K Impressions; No Click-Throughs

And you thought your click-through rate was too low?

Those numbers are true. 60,000 impressions and ZERO click-throughs. But it's by design. Since the middle of March 2006 I've been running a test of Google AdWords.

At the beginning of 2006 I read a lot of Chicken Little stories about online ads and click-through fraud. Especially for AdWords. Of course, Google didn't release any real data. (More recently Google announced click fraud is <1%.) Plus, most of those stories talked about anecdotal evidence while all other stories used WAGs and conjecture to estimate the effect of fraud on ad revenue. I looked at the problem and decided that hard data was missing from the conversation.

I thought about the problem a bit and decided there were three prime economic motivators for AdWords fraud:

1. increase revenue for one or more ad publishers;
2. increase expenses for one or more advertisers;
3. increase Google's revenue.

Those are the only parties economically involved in any click-through transaction. The publisher, the advertiser and the middleman. That's it. Sure the visitor takes part by clicking through, but that person doesn't spend (or earn) any money for that click.

I didn't really want to spend the time to click-through ads on my own site to see how often I could do it without getting flagged by Google. I didn't want to set-up a bot to do it either. I also reckoned that Google would spend a significant amount of effort to prevent such fraud when only a few publishers are targeted. Furthermore, I assumed that Google would announce to the world how it reduced this type of fraud.

For all of those same reasons, I also didn't want to investigate click-through fraud to increase somebody else's expenses. Plus, there was no person or company that I despised enough to ad stalk. Finally, I wouldn't have access to somebody else's records to measure the financial impact. I could talk about numbers of clicks but not hard cash.

That leaves prime economic motivator #3. The conspiracy theorist might say that Google runs bots that click through an ad on each site just a few times a month. It'd be beautiful. Just like taking a penny from the take-a-penny, leave-a-penny jar at a different store everyday. No one site would notice a change in cost. Nor would any individual site see a dramatic effect on revenue. But in aggregate, Google would make a ton of cash. Since Google controls the data and information, it could choose to never release it and nobody would know.

Google doesn't need to do that though. At least not yet.

But what about a computer-savvy person who is also a big stockholder in Google? That person has an economic interest and the capabilities to run the same types of bots to insure an increase in Google's revenue and subsequent increase in stock price. A good conspiracy theory and maybe even plausible.

On top of all of that, I have access to traffic records and the financial impact since I publish the ads. And it's completely passive — I don't have to be involved at all on a day-to-day basis.

So I decided to measure whether anybody is running bots to click through AdWords ads. I put AdWords ads on my site and then played with parameters to mostly hide them. You can't see the ads when you visit unless you look reeeaaally close and know where to look. Of course these ads are always visible to bots.

During this 10-month period (12 Mar 2006 - 12 Jan 2007):

• 45% of my traffic came from organic search at Google;
• 89% of my visitors were new;
• my blog published 59,578 impressions leading to NO click-throughs.

I was hoping for a few click-throughs so I could make a definitive statement that untargeted, bot-driven, click-through fraud does exist. But there was nothing. Nada. Zilch.