First off, this is not traditional media. "Online is widely seen as the most measurable media because every activity whether it's a human or a computer -- visiting a website, viewing an ad, clicking on a link -- is logged by a computer somewhere else," said Dave Osborn, VP-video and media products, Nielsen Online. "Nothing's logging that I'm leafing through the newspaper on the subway."
But offline media at least enjoy established methods of measurement by which everyone navigates. TV, for example, uses one general currency, Nielsen ratings, to plan, buy and report performance. Online, however, uses either ComScore or Nielsen/NetRatings, and in some cases both, to plan ahead of time -- and yet another source, ad-server records, to account for the actual audience reached.
"There's no one way, or one currency, online," said John Sollecito, VP-digital media research, MTV Networks.
Another difference: the fragmentation of a seemingly endless web.
Base of the Long Tail
"In a month, Media Metrix reports on 21,000 entities," said Josh Chasin, ComScore's chief research officer. "And if we get criticized anywhere it's because we don't measure enough, we don't have enough of the Long Tail."
There is one important likeness between digital and traditional that, unfortunately, contributes to confusion over both, Mr. Chasin added. "In the digital world it's easier than ever before to know what machines are doing but those darned consumers are still the fly in the ointment," he said. "It's still hard to know what people are doing."
Given the chaos and unknowns, then, why does every website uniformly insist that its traffic is underreported?
Publishers count web traffic by referring to their own servers, while third-party measurement companies rely on representative panels and sampling. Because these are two very different ways to track audiences, they often result in major discrepancies.
"If you talk to your chief technology officer, he's going to show you the code logs and server sessions that track all sorts of things that may or may not be tracked by outside monitoring sources," said Dean DeBiase, CEO, TNS Media and a former website CEO. "Coming from the publisher side, but also kind of running Switzerland here, I understand both sides."
And neither method is foolproof. Consider the sites using data from their own servers to keep tabs on traffic. Every time a new visitor shows up on a site, the server records it -- but it also installs a cookie on the user so it can recognize return visitors. The trouble is that some people regularly delete their cookies, which leads the servers to mistake those return visitors for additional unique visitors.
A lot also depends on visitors' physical locations in the U.S. and around the globe, typically an important detail for advertisers.
"The Nielsen or ComScore sample is only U.S.-based," said Sherryll Mane, senior VP-industry services at the Interactive Advertising Bureau. "The difference is when those publishers look at their log files, they see tremendous amounts of traffic that are not captured by the samples. In addition, there's an issue with accurately capturing at-work traffic and building accurate at-work samples."
Panels are essential because they get at "the people side of the equation," but they may not be sufficient, said ComScore's Mr. Chasin. "There's been some talk about maybe a hybrid approach is the best solution," he said. "The smart thing to do is to learn about it. We're going to do some work where we work with publishers and put some beacons on their sites; we'll see what we learn."
What to do in the meantime remains a question for all parties. Last spring the IAB publicly pressured Nielsen and ComScore to get their web measurement services audited by the Media Ratings Council; both now have audits under way.
"It's important that we get this figured out pretty soon," said Ned Desmond, president of Time Inc. Interactive, who called measurement an "essential building block" for advertisers and publishers.
Perhaps, though, we should come to grips with web measurement that's slightly less precise than expected.
The internet remains the most measurable medium, ComScore's Mr. Chasin noted. "The consequence is that it ends up being the medium with the most measures," he said. "We need to get comfortable with the notion that commerce can and should proceed with multiple data sets available. We can't let the fact that there's an abundance of riches paralyze us."