Web Rankings Shakeup: It's About Time

Nielsen//NetRatings' focus on time spent on sites, rather than page views, amounts to a sea changeuntil the next best metric comes along

It was the statistic heard around the World Wide Web. Late in 2006, Web-traffic researcher comScore announced that News Corp.'s (NWS) MySpace social networking site had received more monthly page views than Yahoo! (YHOO), long considered the Web's most popular destination site. Suddenly, the media and blogosphere heralded an online sea change. Portals like Yahoo that pull together content from around the Web were on the way out. Social networks were in—and where both people and advertisers needed to be.

Almost as quickly, however, analysts and others began questioning the impact of the comScore (SCOR) study. Should a Web site really be judged by the number of times a user clicks on its pages? What about all the other ways to measure a site's popularity? (See BusinessWeek.com, 12/15/06, "Did MySpace Really Beat Yahoo?").

Shuffled Lineup

Nielsen//NetRatings, comScore's biggest rival in Web measurement, is weighing in. Its verdict? Page views count, but not much. On July 10, the company announced that it would focus on the amount of time spent on a site, rather than pages viewed, when discussing a Web property's ability to engage users. "Time spent is a better denominator than other things," says Scott Ross, director of product marketing at Nielsen//NetRatings.

Talk about a sea change. Using time spent on a site rather than page views rejiggers the lineup of the most popular, or engaging, Web sites from the get-go. Suddenly, the winners are sites that include instant messaging and e-mail features. In May, the top three Web brands are Time Warner's (TWX) AOL Media Network, Yahoo, and Microsoft's (MSFT) MSN/Windows Live. Those nudge aside MySpace and Google (GOOG), which after Yahoo round out the top three by total page views for May.

Even sites that don't make the top 10 by page views register when it comes to total minutes. Apple (AAPL) makes the grade owing to time spent on its iTunes Music Store. As for the Electronic Arts (ERTS) Web site, it finished at No. 8 thanks to the popularity of its free Pogo.com gaming site.

Holistic View

At stake is who gets what slice of the more than $20 billion expected to be spent in the U.S. by online advertisers this year. But first, the industry must figure out which measurement really is the best. Having such a wide range of yardsticks helps marketers better gauge the effectiveness of ads. Instead of relying predominantly on audience projections from companies such as Nielsen's parent, VNU Media Measurement & Information, Web advertisers can ask for stats as specific as how many times an individual user interacted with an ad. But at the same time, the myriad of metrics has created confusion over what advertisers should pay attention to when deciding where to spend that money (see BusinessWeek.com, 10/23/06, "Web Numbers: What's Real?").

Measurement companies do not agree on methods or on the most relevant metrics to indicate certain kinds of users' behavior (see BusinessWeek.com, 4/30/07, "The Travails of Tracking Web Traffic"). For example, comScore believes advertisers must consider page views along with time spent, total visits, and a variety of other measurements to accurately judge the relative worth of competing sites.

It's also beginning to factor in new technologies, known as widgets—sharable applications that often promote products or brands and can be placed on users' Web sites. Andrew Lipsman, a senior analyst at comScore, says the researcher is developing other time-based measurements to be unveiled later this year. "The Internet is always changing fairly rapidly and that is why we tend to take the holistic view of different metrics," he says.

Terms of Engagement

That rapid pace of change is a big reason some metrics are falling out of favor. Page views were once considered a leading indicator of both a site's popularity and, perhaps more important, the amount of advertising space it had to sell.

The reason stems from the way Web sites were designed. Typically, sites could change the content viewed by a user only when the user reloaded a Web page or visited a new one. Advertisers tended to concentrate their dollars on sites with larger page-view counts because it indicated the site's audience was engaged and there was plenty of ad space to sell.

Yet technological advances have erased many of the site-design limitations that made page views a key metric. Using technology such as Asynchronous JavaScript and XML (AJAX), Web sites can change content or the ads on a page without the user needing to reload the entire page. Web sites can also serve ads through multimedia content resting on the site such as in videos or audio podcasts.

Transition Time

Yahoo argued for judging a site by time spent on it after its adoption of AJAX caused it to generate fewer page views and appear less popular in the rankings. Lynn Bolger, Yahoo's vice-president of advertising and sales insight, applauded Nielsen's move. "I think it is a very, very sound decision," says Bolger. "Time spent is more reflective of what the audience is actually doing and how engaged they are in the content."

Of course, time spent doesn't solve all the problems of page views. Time spent does not serve as a proxy for the amount of advertising inventory a site has to offer. A user on a site for five minutes may see one banner ad, or they may see two streaming video ads and several changing banner ads. No one really knows. That means a marketer can only guess at what site has the most available inventory for it to use to reach its audience. It's probably not feasible for marketers to call up every site they're interested in buying on and ask.

Even Nielsen//NetRatings says it's not sure how long time spent should be the prevailing metric. "Right now it is a transition time and, in that transition time, we are going to say that total time spent is best," says Ross. "As time goes on, other norms will evolve."

Before it's here, it's on the Bloomberg Terminal.