Comscore Struggles to Stay Relevant
A week or two ago in the Washington Post there was an article by Josh Chasin, the Chief Research Officer of comScore. The article was describing the challenges faced by marketers in measuring consumers’ consumption of the Internet, and more importantly, Internet advertising.
The article lists the usual suspects as threats to online ad measurement: Ajax, multitasking, rich media, etc. and says that comScore is working on better ways to measure user “engagement” with online advertising.
What it doesn’t list is the threat to comScore’s business model posed by the backlash from websites who say that comScore and Nielsen (the other major online advertising measurement service) do a terrible job of measuring even such basic things as online visitors and pageviews, much less such esoteric things as “engagement.”
As described a little in Mr. Chasin’s column, comScore is moving toward engagement metrics and away from basic metrics such as pageviews and unique visitors. Without actually being an engineer at comScore, it seems that this will rely on extrapolating the online web surfing behavior of a few users to the entire Internet audience, in a similar way to how Nielsen measures TV and Arbitron measures radio.
This begs the question – why? Isn’t the advantage of online advertising that every single action and interaction can be measured? Why then take a new technology that improves greatly on the old ways of doing things and try to shoe-horn it back into the old methods of measurement?
Someday soon, some startup in New York or Silicon Valley is going to figure out an accurate way to measure all web users’ interaction with web content and advertising without violating their privacy. When they do, the existing players are doomed.
January 26th, 2008 at 11:57 am
Here’s the thing. We at comScore aren’t actually measuring the Internet, which is essentially, as senator Ted Stevens so eloquently put it, “a collection of tubes.” What we endeavor to measure is people and the behavior they exhibit. It is people who see ads, look at pages, buy products.
Sure, everything that happens online can be measured and tracked– everything, unforunately, except the last two feet of the funnel. The last two feet is the distance between the screen and the user. Was that click by the same user I saw before, showing up with a new cookie, or is it a new user? Is it a man or a woman? How old is he (or she)? What other websites does s/he visit? How is my ad campaign doing in reaching that consumer? All of these things require knowing that person out there, the last two feet. And none of them can be dealt with “in here,” online inside the most measurable medium. that is why people measurement– which is to say, panels– will remain an indispensible part of the online metrics equasion.
I deal with the research directors at the major publisher websites every day. For the most part these folks understand that Web Analytic data– the site centric metrics that measure every click– are simply not measuring people; at best they measure unique unduplicated cookies. Bots, spiders, cookie deletion, tabbed browsing, US versus international traffic, visitation from multiple computers– all these things serve to drive “Unique” counts generated from site centric data to be 200%, 300% or more higher than the counts of actual different people logging those hits. Happily, Internet media researchers understand this. Even the IAB seems to get it now, as their pending document on reach definitionswill make clear.
Web Analytic (site centric) data and panel data live side by side and both inform business decisions in different ways. But no one says, “I’m switching from comScore to Omniture,” because these two things are two different animals.
Thanks for reading. If you don’t already subscribe to Mediapost’s Online Metrics Insider (where these and other topics get a regular airing), I’d urge you to.
Cheers.
–josh–