Women's site iVillage.com recently hosted a six-week ad campaign for a dieting product from Snapple Beverage Corp. using Tacoda. The Snapple ad was targeted at two groups of visitors, one viewing pages in the iVillage diet and fitness channel and another visiting other areas of the site. Both groups had previously visited the diet and fitness channel at least three times in recent weeks. IVillage said the results showed significantly greater "brand awareness" of Snapple's diet product and an intent to purchase among people shown the ad outside the diet channel. Peter R. Naylor, iVillage's senior vice president for sales, said the results make him think targeted ads could command higher fees. Currently, iVillage charges lower rates for targeted ads than for those run in popular channels.
The Wall Street Journal Online has been using similar software from Revenue Science to show ads based on people's surfing histories since last June, helping advertisers reach audiences interested in travel and leisure activities, among other topics. The site recently did a study that suggests the results have been strong from the new targeting tools, said its vice president of advertising, Randy Kilgore. He added that about 9 percent of ad campaigns on the Wall Street Journal Online use the technology.
Revenue Science's software lets publishers or advertisers type in keywords such as "PDA," "hotspot" and "wireless" to identify groups of people who are likely to be interested in mobile data services.
"We let you search, say, in plain English, 'I want to find technology-savvy, wireless gadget people," said Bill Gossman, chief executive of Revenue Science. "We give you an audience definition, and then score that audience based on the relevance to the targets that you entered in the search box."
Revenue Science and others are starting to use some of the targeting approaches in Web search popularized by Google and Yahoo's Overture division. Google and Yahoo have already taken their targeting beyond search results to serve ads on regular content pages, too, using indexing systems that guess what each page is about and attempt to display relevant messages.
Rather than seeking relevant pages, behavioral targeting attempts to find relevant people based on what they've read in the past. In the past two months, a string of new products have been announced. In March, 24/7 RealMedia Inc. said it would offer advertisers the ability to serve behaviorally targeted messages across a network of 700 participating sites. Last week, aQuantive Inc., another ad network, said it was forming a division to offer behavioral targeting, too. AQuantive manager Scott Howe said his company's system collects data about site visitors in broader categories -- about 30 of them -- than just what they read, including what they've bought from advertisers and how they've responded to particular ads in the past.
Some publishers remain leery of allowing data about their audiences to be shared with data about visitors of other sites in large networks. "I wouldn't be comfortable with that," the Wall Street Journal's Kilgore said of ad targeting that lumps rival Web sites in one networked system.
But the software providers are hoping publishers will get over their objections if they find they can collect more money when advertisers aggregate audiences across multiple publishers.
Greg Stuart, chief executive of the Interactive Advertising Bureau, said he thinks behavioral targeting is still in its infancy, but he said there is no question it will prove more efficient at delivering ads than the demographic categories driving most advertising today.
"Demographics are a horrible predictor of behavior," he said, "and some $250 billion in advertising is essentially based on that today."
