If you’ve had the feeling lately that you’re being followed as you read the Wall Street Journal online, well, then, you might have ESP. Beginning in April, WSJ.com implemented Audience Search, a new service from Revenue Science Inc., a data-mining firm coheadquartered in Seattle and New York City. Dow Jones says that Audience Search has improved matches between its readers’ interests and its advertisers by as much as 218 percent.
In the past, WSJ.com, like many content sites, served up click-through ad banners based on the section or column a reader selected. The innovation of Audience Search: It looks for keywords within a selected article, and ranks that article’s relevance to a particular topic based on the prominence and frequency of those keywords. Audience Search also groups stories with similar keywords. At the same time, when users visit these stories, they too get tagged, “followed,” and ranked, depending on the number of pages with similar topics they visit. Combine these two data sets, Revenue Science maintains, and you have a better-qualified set of eyeballs, or leads, to serve to advertisers. But—and this is the crux for privacy advocates—this is all accomplished without knowing the identity of the actual user.
Revenue Science is so confident its system works that it has implemented a fee structure similar to that used by rich-media technology providers: The firm doesn’t charge publishers at all, but gets paid by advertisers who are charged a fixed CPM (cost per thousand) rate.