Numbers game

The Times goes on an odd attack against and its claims to be an audience leader.

The Times’ real complaint is against Comscore and other providers of the numbers — like every big web site — uses. These numbers are gathered from small samples — a la Nielsen TV ratings — and they are relative bullshit but they’re what advertisers go by. Their veracity is ever less reliable the more the media world spreads. When there were only three networks, a small sample was probably a fairly reliable indicator if not measurement of viewership. But with millions of sites, the odds that a small sample will go to them in the same proportion as the rest of the world falls to nada. (This, by the way, is why blogs are doomed in a Comscore/Nielsen world; there’s no way that they can be measured. And that, again, is why we have to do a better job with our own measurement.)

This argument over ratings numbers is less important online because advertisers need not care how big the site is, only how often the ads they pay for get served — something the web can measure and verify and TV can’t. I served on the horrible Audit Burea of Circulations committee that dealt with these issues years ago (“What is a pageview?”). As it turned out, advertisers didn’t care about audits of how big sites were. They needed audits of their own ads. So whether Forbes has X million users or half that, it doesn’t matter to the advertiser so long as his ads get served to the right number of people.

One more note: I think I found one odd reason why keeps growing. For unknown reasons. GoogleNews favors way over other sites. Look at this analysis of citations on GoogleNewscompiled for more than a year since GoogleNews had a neonazi site and wouldn’t reveal its own sourcdes — and note how high is. The guys didn’t even know this until I ran into them on a panel and told them.

: Rafat Ali weighs in.

  • What is odd about the NYT attacking a capitalist icon?

  • Forbes has been telling advertisers that they are attracting Business decision makers, but in reality it’s the Mom what’s for diner demographic that’s doing the clicking.

  • Joe

    Having valid audience measurement techniques are important, but the only ones that are truly accurate (Javascript “beacons” embedded in pages) are the ones that advertisers don’t trust. Nielsen, Comscore, Hitwise, and their ilk are notoriously inaccurate, particularly for small sites. We have frequently compared their figures to our actual numbers, and found them off sometimes as much as a factor of four.

    But developing good and externally verifiable audience measure techniques is essential. Jarvis misses the point (“advertisers need not care how big the site is, only how often the ads they pay for get served”). A company isn’t an advertiser until it buys an ad. And why would someone buy an ad on a site? Because it reaches a critical mass of the desired audience. So once someone is running the ad, they get fulfillment information from DoubleClick or other ad servers, so Nielsen becomes irrelevant at that point of the relationship.

    If ABC or someone could figure out a way to audit a site’s Omniture/SiteCensus/Urchin/whatever set-up on a regular basis, we’d then have great audience numbers for major sites. Sure, the little, itty-bitty sites won’t have audience, but advertisers aren’t interested in them anyhow.