OK, it's unlikely the two web-measurement giants will be unseated anytime soon, but unlike in other media, where multiple measurement companies have eventually been whittled down to one dominant player, some advertisers are cautiously welcoming the multiple audience-measurement options available online.
"Where ComScore and Nielsen fall down particularly is in tracking small sites," said Harry Case, director-analytics, MindShare Interaction. He cites the example of a pharmaceutical marketer looking for condition-specific sites: "Often only two or three condition-specific sites are picked up by the syndicated research services [Nielsen/NetRatings and ComScore]. If you're talking about trying to put together a plan for high blood pressure, it is almost impossible to understand the demographics of those sites, so we have to resort to other information."
Not an 'insurmountable challenge'
The service he has looked at most closely is Quantcast, which places a pixel or "tag" on publisher sites and then keeps a record of those viewing the site's content. "As media choices increase and a panel stays the same, the panel's ability to resolve the details diminishes quite quickly," said co-founder and CEO Konrad Feldman. He admitted that Quantcast relies on placing cookies on computers -- which ComScore in particular has argued leads to inaccuracies: Cookie deletion leads to overcounting unique visitors. But Mr. Feldman said "that doesn't mean it's an insurmountable challenge."
Compete was born out of IdeaLab, a technology incubator that also has spawned companies such as Citysearch and Picasa. The service has been around since 2000 but recently has taken on a more proactive role in marketing its services to advertisers. It's not so much focused on the true Long Tail sites of the web as on what Compete's Stephen DiMarco calls the torso, or the top 500,000 sites. After that, he said, audience falls off fairly quickly. The company buys the rights to use ISP data and supplements that with a panel of two million.
Of course, the accuracy of such services is a less-agreed-upon issue. The Amazon-owned Alexa service, which is commonly used by reporters and bloggers because of its ease of use and availability, is often targeted for inaccuracies because of what people call a self-selection bias. Its measurements are culled from the web travels of those who have installed the Alexa toolbar, and because the toolbar is particularly useful for (and, thus, most likely to be installed by) webmasters and search-engine optimizers, it tends to overweight websites popular with that crowd. Alexa didn't return calls, but its website claims it has "several million" toolbar users and admits that certain sites may be undercounted (it doesn't work with the Opera browser, for example) or overcounted.
A big challenge, said MindShare Interaction's Mr. Case, is that there's no standard against which marketers can measure the accuracy of the new services. For the moment Nielsen and ComScore are the gold standard, and the best marketers can do is benchmark new services against them.
How do all these free services make money? Right now, venture-backed Quantcast doesn't. But it's evaluating multiple revenue models centered on becoming an intermediary that helps smaller publishers increase the value of their inventory. Compete charges clients for consulting services, said Mr. DiMarco, who lists Hyundai, DaimlerChrysler and Sprint as clients of the service.
"You can have one tool and static [measurement services] or multiple tools and innovation," said Andy Fisher, director-analytics at Avenue A/ Razorfish, who points out that the barrier to entry is also lower in online measurement than offline. "A healthy market pushes innovation."