More on why I think time spent on site is a kind of a dumb metric

So now that time spent on site is the new belle of the metrics ball, I just want to register my disapproval again. It’s a great and interesting number, for sure, if you could actually measure it accurately. As far as I know, though, you can’t. I don’t want to go into a lot of detail again on why this is, please check out my last post on the subject where I ramble on at length about why that happens to be the case.

This post is about shining a little more light on the fact that it is a real world issue. Two companies, well, one company now, that I respect both give me a time spent on site metric. Feedburner and Google. I don’t think anyone will question that these companies are run by smart people that are going to do a bang up job pulling numbers out of your traffic. And it’s true, the visitor numbers they pull differ but maybe by a percent, maybe 2 or 3 on a bad day - that’s all margin of error type stuff. Very reassuring that they match up, too, let me tell you.

With visitor data so close, it is reasonable to assume they are recording the same bits of traffic, so even if they are wrong, they are wrong over the same data. They both also offer an average time on site and I think it’s illuminating to check them out.

Feedburner gives me a pretty uniform range of between a minute or two. People come to my site, check out a page or two and jam on. Sad, true, but a fact of life. Sure some people check out more pages and some may spend more time reading the sheer genius that lies within my posts about remote control planes, but many come read what they were searching for and move on.

Google Analytics, on the other hand, has a huge variation in average time. It’s anywhere from 4 minutes up to, I think I saw 15 minutes. That’s average time per user. Google thinks that one day people spent an average of 15 minutes on my site - and I don’t have such low traffic that a few people could skew things that much. It simply is not within the realm of conception.

These two companies are grabbing the same traffic, looking at the same raw data and coming to wildly different conclusions, and that’s what I was talking about in my last post. Since you simply can not know when someone is looking at your page and not stashing your page in a tab, or loading it up and then going to home for the evening, the choices you make in how to calculate this number has far more effect on the resulting time than however much time people actually are spending on your site.

The problem is, advertisers don’t actually care about this technicality. They want a number that in theory gives them information on something - if someone like Nielsen gives it to them, then they’ll all believe it. As long as they are all believing the same lie then nobody’s getting an advantage, they’re all swimming in the same piss pool. Meanwhile the still perfectly valid (for the majority of sites on the internet) and much more correctly measurable stats like visitor sessions gathers dust.

Jul 24th, 2007 · Comments RSS · · · · · Tags: analytics · google