But while the move by Yahoo brings a new level of accountability to the Web, it also points to weaknesses in how ad impressions have been measured until now.
"There are double-digit percentages of inventory that may not today be counted because of the guidelines," said Jonathan Adams, VP-media director at Modem Media, who worked on the guidelines with IAB. "By shifting to a more conservative counting methodology, top publishers are forgoing millions of dollars in advertising they would have been able to sell under the old counting methodology."
Yahoo's new plan is to count an ad impression only after the full ad has appeared on a user's screen, and the user's browser has sent code saying the ad has been fully rendered.
It goes a step further than a set of guidelines recently adopted by the Interactive Advertising Bureau. Those guidelines indicate that an ad cannot be counted until the user's browser sends a message requesting the ad. Yahoo's system waits for the full ad to appear.
"Yahoo has taken an additional step of moving to the very, very conservative end of the standard," Todd Teresi, Yahoo's VP-global sales operations, said. (Mr. Teresi chaired the IAB committee that established the guidelines.) Other companies following IAB guidelines include CNET Networks, Weather.com and Univision.
Ad servers, which include companies such as DoubleClick or Atlas, report an ad as having been delivered as soon as they are asked to send out the ad to a user's desktop. But there are a number of reasons the ad might not be delivered, such as ad-blocking software that keeps the ad from being seen by the consumer. Now, Yahoo says marketers won't be charged for those blocked ads. Web ad rates are often determined by how many times an ad is viewed.
Yahoo's method of counting will cut third-party ad servers out of the process and leave the publisher as a marketer's only source for guaranteeing ad impressions. IAB created the guidelines to reconcile discrepancies between what a publisher says is the number of impressions served and an ad server's count.
"If only the publisher is counting, then it's difficult to have a checks and balances," said Jeff Marshall, senior VP-managing director of Starcom IP. "Most agencies use third-party ad servers across sites in a given campaign to be able to compare apples to apples."
Several circumstances can contribute to an ad server's count and a publisher's count not matching. Sometimes an ad server doesn't get the piece of code that requested the ad. And sometimes the user sitting at a PC exited the Web window or hit stop before the ad was loaded. Discrepancies often vary widely, in some cases by as much as 50%, Mr. Adams said. In those instances, the ad-serving firm and the publisher would have to reconcile the numbers and reach a compromise.
Conflicts were so commonplace that the industry typically didn't investigate any claims below 10%, executives said.
Being first to announce ad-counting changes even when the entire industry will follow suit does seem a bit self-serving, other online ad pros said. "It's a differentiator," said Renny Gleeson, managing director, Carat Interactive. "It reinforces their position as a market leader."