Web Measurement Morass
Earlier this summer, comScore Media Metrix, one of the two leading providers of third-party Internet traffic data (along with Nielsen NetRatings), dramatically revised its numbers for Forbes.com, which had been claiming a worldwide audience of 15 million unique visitors, based on a monthly average. The new numbers showed 13.2 million visitors in February and 7.3 million visitors in July.
It's a common problem around the industry. Web sites claim explosive growth but there's no real standard for what to report or how to report it. Neither marketers nor publishers are clear on what terms to use or even what's important and what's not. And ultimately, in the "go-go" atmosphere of Web 2.0, not too many are worried about the gray areas of online traffic reports, at least not yet. "A key issue is the discrepancy that has existed since the dawn of Web advertising, and persists to this day, between third-party ad server logs and the Web site's own traffic logs," says Scott Karp, managing director of research and strategy at Atlantic Media and publisher of the Publishing 2.0 blog. "Since everyone has grown up with data discrepancies online;and because through the dot-com bust there wasn't a lot of money at stake, there is generally acceptance that problems with the data still exist."
That leaves publishers and advertisers doing the best they can with what they have. "I don't think anyone is happy with Web measurement but there's no good middle ground," says Joe Galarneau, chief technology officer at New York, who adds that Nielsen NetRatings typically underreports his Web site by 50 to 75 percent. "If you had a site-based measuring system and a third-party auditor like an ABC, then you could run your metrics and get audited for the results. You'd have something that's verified, not some BS fairy tale panel-based stuff. We want that and advertisers want that but we're not really seeing any great hue or cry over it yet."
What to Look for
The traditional metrics remain the most widely accepted: clicks and unique visitors. But that may not be the best route for publishers. "It depends on their objective," says Eric Shanfelt, vice president of eMedia strategy at Penton Media. "Certainly, page views per month is critical. That's the most concrete measurement of a Web site's overall traffic numbers and suggests how many given ad impressions you'll be able to serve. I like to look at unique visitors per month and page views per visitor."
But when most sites say "unique visitors," they're really referring to total visits. Even audience research and auditing company I/Pro's own audit statement reports page views and visits per month. "The unique visitor is the closest thing to a print magazine circulation," says Shanfelt.
He adds that publishers also need to be aware of distinguishing non-human traffic such as bots and search engine spiders. "There's an interest in things that don't rely on log analysis and can filter out non-human traffic," Shanfelt says. "I've seen so many other b-to-b Web sites where they're not filtering out automated traffic."
New York looks at content conversion rates and path analysis, such as how is navigation working in terms of people going from screen A to screen B, and how times visitors access a certain page.
Galarneau also advises publishers to pay closer attention to search engine traffic to help develop search engine optimization strategies. His magazine uses a service from WebTrends that sits on its site and tracks search engine activity. "It takes a little bit of setup but once you do it, it can run every day," says Galarneau. "A lot of these measurement packages give you real-time information, others update a few times per day. If we get picked up on Yahoo for a story, it lets us know immediately so we can accommodate a big spike in traffic. If you're falling down in certain areas you can look at how you're tagging pages."
Most discussions of online advertising make it sound like money is falling from the sky. However, the lack of measurability will have an increasing effect and publishers will need to be prepared as advertisers grow more sophisticated. "I just talked to an advertiser who's very hesitant to get into online advertising more heavily," says Shanfelt. "They're putting more dollars into direct marketing instead. That's an extreme example but it's happening."
Both advertisers and publishers need to start talking site metrics, rather than traditional ad metrics. "Tech advertisers are usually very sophisticated but other markets are looking at the ad position, clickthrough rates, things that aren't necessarily site metrics but ad metrics," adds Shanfelt. "In b-to-b, we have a way to go in many markets to truly quantify what we have and then quantifying what an advertiser gets when they spend their money with us. There are still many publishers saying, 'Do you want to rotate a banner on my site, it's $500 or $1,000 per month.' But you don't know if there's one person rotating, 10 people rotating, 20 people rotating. We still have to do a better job as an industry of communicating the exact value of what someone is buying from us."
New Media, New Metrics
Part of the problem with Web metrics is that the capabilities of the medium are changing so fast. "I hear a lot about clicks and uniques going away and they probably are," says Galarneau. "Rich media is driving that movement away from clicks. Applications like Ajax let you hold your mouse over something and another screen pops up. You're still on one page but you're getting multi-pages' worth right there. Podcasts are an issue too. When you download once and share with friends, how does that translate to off-line media?"
Galarneau says that he's starting to follow metrics like abandon rates but it's hard to know what to look for. "This is such a new area, no one knows what's a good number or what's a bad number," he says. "People put stuff out there and see what happens."
Both publishers and advertisers seem to feel that there's no use getting too worked up over existing measurement metrics, faulty or not, when there's no clear consensus on what the medium may ultimately deliver anyway. It could require completely new metrics.
"When you consider new trends like behavioral targeting, which are delivering advertising to people, not Web pages, then add in search advertising, which doesn't care about Web measurement either, you can see why the Forbes.com story died so quickly," says Karp. "It's not quite clear that gross audience measurement is the future of online ad targeting, so nobody's sure whether it's worth getting upset over problems with the data. If you can precision-target individuals online, gross audience measurement may turn out to be an artifact of old media."