Yahoo: Lower Page Views Due to AJAX

Yesterday's comScore Media Metrix report, exclaiming that estimates of page views for social networking site MySpace at 38.7 million for the month of November topped those for Yahoo by 600,000, came with some small print that should have generated at least an asterisk beside today's headlines: Yahoo has recently moved to an AJAX-driven page model, which probably reduced the number of complete page refreshes per user.

Web analytics software used by comScore and other services treat "page views" as complete refreshes of an entire page. But Web sites that use more modern approaches to layout have recently adopted Asynchronous JavaScript to enable browsers to refresh select portions of the page when necessary, reducing bandwidth and improving layout. Yahoo is one of the more recent inductees to the roll of sites embracing this technology.

Along with those efficiency gains comes the headache of what kind of a page view counts as an "impression" for the sake of advertisers, for whom Yahoo's page view numbers have the biggest impact. ComScore's estimate is that Yahoo's page views declined by 3.5 million in a one month period alone, even though experts in the field of Web design have been arguing since last spring that deploying AJAX reduces the efficiency of analytics software.

That didn't stop Fox Interactive, the parent of MySpace, from touting what it calls 200% year-over-year growth, as evidenced mainly by page view estimates from comScore Media Metrix and Nielsen/NetRatings. Experts point out that MySpace does not use AJAX.

In a May 2006 article for Clickz.com, managing editor Pamela Parker openly predicted that Yahoo's AJAX-enabled home page rollout - which was in beta at that time - would directly lead to reductions in page view count, though not necessarily in real traffic. One example of an AJAX control Parker noted was a "personal assistant," which works like a mini-browser, serving up different small batches of content from throughout Yahoo's mix of services.

Yahoo officials told Parker for that article that, while they planned to attach a fix advertising space to that AJAX control, which would remain in place while users continued to browse, they had no model in mind for how they would count advertising impressions for that control.

"I'm certainly not suggesting we halt progress in the interest of simplicity," Parker concluded in May. "I'm just saying things are getting more complex every day, and it's time to start asking about more than just page views, about more than just impressions."

Last September, the Internet Advertising Bureau urged both major Web analytics companies to consider reforming its analytics measurement processes to account for partial refreshes, or to introduce new means for measuring advertising impressions. IAB cited real-world stories from major sites such as MSNBC.com, which noted strangely instantaneous drops in its own recorded page views by as much as 40%, after new layout measures were adopted there.

The following month, a BusinessWeek article told the story of instant messaging service provider Meebo, Inc., whose recorded page views omitted huge classes of extra traffic, including referrals back to its home page from a link embedded in its IM device. The company's proprietor invited comScore officials to look over the data first-hand, and while they obliged, they couldn't come to any conclusions.

The possibility that analytics measurement schemes could be months, if not years, behind the times could actually be good news for sites such as CNET, whose stock value and even corporate status have come under fire recently in the face of page view numbers that appeared to decline by 55% just between September 2005 and September 2006. During that period, CNET instituted a radical simplification of its Web page structure that includes the use of tools such as Asynchronous JavaScript. CNET's analytics service is comScore.

In an open letter last September, comScore President and CEO Dr. Magid Abraham responded to criticisms posted in MediaWeek and AdWeek magazines, that its page view figures weren't matching up with Web servers' own logs. "When you scrutinize the details," Dr. Abraham wrote, "the answer to the question about why Web log and panel-based data don't always match up is...'it depends.'

"In fact, the reasons for discrepancies depend on a number of factors: the panel data could be wrong, the Web log data could be wrong - or more often - they are both right given the exact definition of what they measure," Dr. Abraham continued. "But, those definitions could be vastly different." Later, in a section of the letter entitled, "Are We Even in the Same Universe?" he added, "Technologies like AJAX are making it increasingly challenging to even define a page view and will increasingly become a potential source of discrepancy."

For his personal blog, Steve Rubel, the senior vice president of public relations firm Edelman - one of the world's largest - predicts the page view as we know it has four years to live, tops.

"This is a dirty little secret in the advertising business that no one wants to talk about," Rubel writes. "Media companies love to promote how many page views their properties get. They've used the data to build equity. They will fight it tooth and nail to protect it, perhaps by not embracing interactive technologies as quickly as they should. But that's not going to stop the revolution from coming."

© 1998-2014 BetaNews, Inc. All Rights Reserved. Privacy Policy.