Tags for measuring click-to-page dropoff?/
July 29, 2006
A couple of weeks ago, a publisher asked for help. His advertisers were upset because they weren’t getting the kind of clicks that he was reporting. I’d like to put a campaign code on the links, he said. Is there a universal campaign variable that I can use — one that will work with anyone’s web analytics?
Normally, I wouldn’t have gotten involved but my friend Ian Houston asked me if I would write the publisher. (I can’t imagine why, he is a brilliant technical guy.) Since my expertise is in marketing, I succeeded in getting completely tangled up in the technical issues: I started to study the problem of which major WA packages could all accept the same campaign variables and which, like Google Analytics, had to have specific tracking code.
After I was done, I went up to 30,000 feet and wondered — did it really matter? After all, if advertisers are seeing fewer clicks than the publisher is showing, having them nicely tagged isn’t going to change anything, now will it? The advertiers won’t see more clicks, just pretty web analytics, right?
So I wrote Dylan Lewis of Intuit. Dylan wrote Hack #51 in Web Site Measurement Hacks, calculating click-to-visit drop off. He agreed that tagging won’t make clicks show up, but offered these pearls of wisdom:
Without making sure that everything is tagged, you will never properly diagnose the discrepancy because the argument of “not everything is tagged” will trump both the person serving the ad and the person serving the landing page.
Separately, Ian added at least one other reason that the effort was worth undertaking:
At least with the query string method they can use thresholds and averages to see when something truly unrealistic is going on that needs to be looked into.
So there you go — another lesson in web analytics that is completely unintelligible if you don’t understand campaign codes. (So if you have to learn only one thing about your web analytics, learn how to tag links with campaign code. It’s a great investment of time.)