• by Tig Tillinghast , May 16, 2002

Last week, WebSideStory, a company that analyses traffic data, put out a study that seems to prove what we’ve long suspected – that some sites’ methodologies for counting traffic have still more holes in them. In particular, they harped on sites that use the “meta refresh” redirect and fail to account for the extra page views that show up in server logs.

The company reports that the mistake can lead to traffic inflation approaching 30%.

This gave me a flashback to the mid-90s, when companies with security software to sell would commission “studies” to show how insecure our e-commerce really was. The net effect of this, of course, was to create panic and trepidation among people who otherwise would have been buying things online much earlier.

Headlines would read, “Online buying unsafe – until now…,” but then another headline the next week would read, “Only now can secure transactions remain private.” The impression the viewers began to develop was quite different: “Internet guys are a bunch of bozos.”

Except this time around, it’s a little different. The online advertising numbers really are polluted. In fact, they’re often different enough from reality that it prevents marketers from getting the resolution of accurate data that is the very thing that makes the Internet worthwhile for them.

The same “bunch of bozos” impression is being created around the Internet medium, but this time it seems to be well-deserved.

I’ve heard people argue that people like me worry too much about this sort of data issues; that it doesn’t really matter what metric we use for online media, as long as we pick just one. That way, we can tell which sites are being rewarded with dollars for what amount of media.

This is sort of an economist’s point of view. If everything were averaged out, all we’d need to know is the gross amount of media – no matter how consistently inflated or deflated. And this makes sense to an extent. With a limited supply of dollars chasing the impressions out there, theoretically those same dollars would reward the same sites in the same proportions, no matter which definition of an impression we chose to standardize.

It could be measured by the truckload, in parsecs, cup sizes or megawatts for all they care. When everything’s averaged out, all they need to show is the proportion of spending and the proportion of media allocated.

But to the marketer wanting to see the medium as a new science, a set of rules, laws and levers with which discovery can give your company a significant competitive advantage, these measures simply become irrelevant. These marketers want one thing: the truth. They want to know who saw what, under what circumstances. Only then can they guess at why those viewers did what they did subsequently. Everything else is sort of like the Gross Domestic Product: it’s a generalization that seems to suggest a trend, but it has no relevance to any individual company.

So, despite the fact that their study adds to the long annals of self-serving “research,” bully for WebSideStory. May they continue to point an accusatory finger at the naked emperor sitting on the throne of online advertising data.