Many thanks to Shuman Ghosemajumder at Google, who finally got back to me on my continuing saga of MyBlogLog and AdSense.Shuman and the Google engineers have taken an in-depth look at my stats (including referrers and browser stats) and have come up with an explanation. The scenario described draws attention to a potential flaw in all click-tracking solutions, and (according to the engineers) is partly due to the fact I’ve been getting a lot of traffic from StumbleUpon.com.
The way that MyBlogLog tracks ad clicks for Firefox users is to hook thepage unload event and record the position of the mouse on every mouse event.On page unload if the last mouse position was over an adsense iframe it iscounted as a click. The ad placement on the Chocablog site is at the topof the page right under the Stumbleupon toolbar, so it’s likely that whenpeople go to click “stumble” their last mouse event will be over the adiframe and a false click will be recorded by MyBlogLog.It seems like whether the user is using Stumbleupon or not, many of theseclick tracking scripts will be vulnerable to false positives when the adframe is at the top of the page and users leave via the back button or theirfavorites bar.
Basically, the only way of tracking clicks on Firefox (and thus, the method adopted by pretty much all click-tracking software) is to look at the last position of the mouse pointer when a user leaves the page.Because StumbleUpon users use the toolbar to ‘stumble’ to the next site, and the affected site features a leaderboard of AdSense ads at the top of the page, people often move the mouse over the ad block before leaving the site.Convinced?This does sound like a logical explanation to me, and if true, shows up a major problem with tracking clicks.But I still think there’s more to it than this. After being featured in B3ta’s newsletter last week, the same page saw another spike in traffic, so I had another opportunity to look at what was going on. This spike was accompanied by an increase in clicks on other links and ads (albeit a small one), but the number of clicks reported in my AdSense stats actually dropped.I’m unsure why users would click other ads and links, but seemingly avoid AdSense ads – including the most prominent ad-block on the site (the top-of-page leaderboard).So while I think Shuman’s explanation seems quite plausible, I don’t think it’s the full story. But somehow I doubt whether I’ll ever find out what’s really going on.