And this spring the director of strategy and insights at MEC Interaction is finally getting some, as clients AT&T Mobility and Monster.com engage in a project called engagement mapping, from Microsoft Corp.'s ad-serving arm, Atlas. The project aims to weigh the influence of each ad in a campaign, rather than rely solely on clicks to measure success.
"We plan to evaluate the reporting in April, and testing could go through May," Ms. Zola said of the trial, which underscores one of the great paradoxes of online advertising: Everyone calls the internet the most measurable channel, but that claim hinges on whether marketers are really measuring the right things -- and increasingly, it appears, they're not.
The industry grew up around a "last ad" measurement model, which meant the credit for a consumer conversion went to the last ad a consumer clicked or viewed before the conversion. A byproduct of that was the ads a consumer saw leading up to that "last ad" were deemed worthless.
New success metrics take a more holistic look at all the ads a consumer encounters along the path to conversion.
Atlas isn't the only company working on a new model. The metrics, in various stages of development at a host of ad-serving and media planning and buying firms, aim to become the new standard for measuring the performance of advertising and marketing campaigns, and possibly for pricing ad inventory.
Taking into account display, rich media and other tactics, along with the last ad clicked, media planners can "give credit where credit is due," said Emma Pop, associate research director-insight and analytics at Starcom USA, which last year started a similar service, Multiple Attribution Protocol, through an application called Artemis.
For example, a consumer with a Yahoo e-mail account also uses Yahoo Finance. She sees several ads for investment companies in both those places, but it's not until she searches on Google that she clicks on an ad. The engagement-mapping approach asserts that those Yahoo ads had value in their influence and should share in the credit that traditional models would give solely to the search ad.
There's evidence that the impressions leading up to a click matter. Microsoft published research last year that looks at interaction between search and display advertising. The conversion rate on average was 22% higher for consumers who saw display advertising for a company before clicking on that specific company's search ad, compared with those who didn't see a display ad. Some ad clients experienced as much as an 80% to 90% uptick.
"One of the things we're seeing as we go through the beta process is video performance improves under this model," said John Chandler-Pepelnjak, Atlas principal analyst. That's because rarely does video or rich media become the last click. "If you are Best Buy and you're buying video [advertising], you'll probably see a change in the metrics for video performance from being one of the worst, in terms of cost per action, to much better. This might prompt you to increase your spend in video or at least not take it off the buy."
Leaving a trail
The concept appears simple, but the technology is complex: raw log-file data, time-stamped and collected by ad-serving companies like Google's DoubleClick and Microsoft's Atlas, along with a short line of code known as a pixel hidden in web pages, keep a record of each time consumers enter or exit a web page, click on a link or ad and enter information in a search box or application. Those data are fed into software platforms designed by companies such as Atlas, Epic Advertising, Media Contacts and Starcom.
"It's sort of like reading an advertising diary," said Ben Winkler, VP-interactive media director at New York-based Ingenuity Media Group, which joined Atlas' project earlier this year. "It's like you opened a diary where someone wrote, 'I saw three billboards, I heard a radio ad, saw a few banners ads, and searched through Google to find and buy the product.' "
Rather than wait for a crisis to tell the advertising client something isn't working, media buyers can rely on these data to identify when consumers had contact with the ads, even if it's an hour, day or week later.
"We know the person saw ad No. 4 on Yahoo Finance an hour ago," said David L. Smith, CEO at Mediasmith, which is participating in Atlas' and DoubleClick's tests with advertisers. "Embedded code in the pixels lets us track the pages and things they interact with on the site."
Ideally, the model would also take into consideration billboards, newspapers, magazines, radio and TV, along with about 20 other types of media not found online, and the next step in engagement mapping will be to pull in some of those media. For example, how did the Super Bowl ad from General Motors Corp. affect the consumer's online buying decision?
"We see there's a solid 30% lift in conversions when people are exposed to online advertising outside of search," said Rob Griffin, senior VP-U.S. director of search, data and analytics at Media Contacts. "We see a 200% to 300% lift from people searching for particular brands when they're exposed to offline and TV advertising."
Marketing company Brand Keys measures clients' expectation on the basis of 26 media-buying options and allocates a percentage to each. "You can visit and click all you want, but unless you track offline media and actual buying, how do you know the campaign is successful?" noted Robert Passikoff, founder and president.
Similarly, through statistical analysis Ingenuity Media Group media buyers can see cause and effect between media investments, impressions and actions taken by consumers either in stores or online, Mr. Winkler said, but the industry is far from making it work because "we still don't know if Joe drove by the billboard today."