The search for reliable measure of Web ads

Online advertising jumped 25 percent this year, raking in a cool $20 billion, but Internet executives say that figure could have been even higher if advertisers had reliable and consistent ways to measure online audiences.

Unlike traditional media, where each format has one main ratings provider - the Nielsen Co. B.V. for television, Arbitron Inc. for radio and so on - there are many sources of data on online audiences. And they frequently conflict.

Disagreement also continues over which criteria best gauge users' potential interest in a product or service. And the resulting data are not easily comparable to ratings in other media anyway.

It is a "problem of plenty," as Manish Bhatia, president of global services for Nielsen Online, a unit of the Nielsen Co., said at a recent conference on measuring online audiences.

Web publishers are frustrated that the lack of cohesion is holding them back from capturing more of the $250-billion-a-year U.S. advertising pie, especially given the huge amount of time people spend online.

"This industry looks like it can't get out of its own way," said Steve Wadsworth, president of the Walt Disney Co.'s Internet group. "We need measurement of the audience and their use of the system that's clear, simple, and actionable for a marketer. You need comparability with other media."

As Internet executives hash over clickstreams, page views and user panels, 2008 is sure to see even more evolution of the way online audiences are measured. Other media - including TV, radio and billboards - also are revamping the way they calculate ratings in response to pressure from advertisers trying to measure how effective their ad dollars are.

David Hallerman, senior analyst at research company eMarketer Inc., said many large advertisers stayed away from the Internet because of confusion over audience measures. Some also want to stick with video ads, which are still in their early stages on the Internet.

The Interactive Advertising Bureau, which represents more than 300 Web publishers, has called for Nielsen Online and comScore Media Metrix to undergo audits by the Media Rating Council, a process that is still under way. ComScore and Nielsen both still use panels, while Quantcast Corp., a relatively new agency, combines panel and Web-based data to produce ratings.

Resolving what to measure is as complex as deciding how to measure it. Some sites produce their own ratings based on internal server logs, on the theory that panel-based data understate traffic. But comScore said internal logs could overstate traffic when users delete identifying files called cookies from their browsers, because servers think they are seeing a "unique visitor" each time that user arrives.

Counting unique visitors can also be challenging - and lose meaning - when an individual logs in to several different computers, or a family of six all use the same computer. "Page views," once a key indicator, have not been since Ajax software let people view different elements on one page instead of going to a new page for each one.

From any vantage point, there is still no clear equivalent for reaching a potential audience of 18 million people around the country at the same time with a single ad on

Desperate Housewives.

"There aren't well-established, tried-and-true standards in the industry, which need to be worked through," said Jeff Marshall, senior vice president of digital marketing at Starcom USA Inc., a major ad-buying agency.

Traditional measures may not even apply to the Web, some executives say, because the benefits the Web offers - most notably, the opportunity for users to click right through and buy the advertiser's product - are not comparable with other media.