By Peter Whoriskey
Washington Post Staff Writer
Thursday, May 22, 2008
How do you find a bride these days?
One of the nation's leading online tracking companies knows.
Monitoring consumers at roughly 3,000 Web sites, Revenue Science identified brides by picking out bridal behavior it had seen: anyone who'd gone online to read about weddings in the news, entered "bridesmaid dresses" into a search engine or surfed fashion pages for wedding styles.
The company found 40,000 such people, whom it knows by random number, not name, and sent them a tailored online ad.
"A successful campaign," according to company president Jeff Hirsch.
The growing practice of "behavioral targeting," or sending ads to online users based on their Internet habits, is now under scrutiny by the Federal Trade Commission, whose review could shape not only Web advertising rules but the character of the Web itself.
For while public interest groups argue that compiling profiles of largely unsuspecting Internet users ought to be illegal, online advertisers and publishers respond that their ad targeting tactics protect privacy and may be essential to support the free content on the Web.
Behavioral targeting allows many Web sites to raise ad prices, because advertisers will pay more when they can isolate a particular audience.
Limiting behavioral targeting could "jeopardize the consumer's ability to get free content on the Internet," said Paul Boyle of the Newspaper Association of America, a trade group that represents the business interests of most U.S. dailies, including The Washington Post.
The FTC is considering guidelines, for now voluntary, that would make it harder to target behavior. The principles were issued in December after town hall meetings, and the public comment period ended last month.
As the commission's deliberations begin, some federal and state lawmakers are weighing measures that would be mandatory. New York lawmakers, for example, are considering a law similar to the FTC guidelines.
Now that many Americans spend as much time interacting with the Web as they do watching TV, there is a wealth of information available for targeters: what articles a person reads in online newspapers, what queries he or she types into search engines and what items the person shops for.
Revenue Science and its peers say that because the user profiles they keep are organized by randomly assigned numbers, no personally identifying information is ever stored.
But privacy groups argue that while the items collected by targeters may be "anonymous" when viewed individually, taken together they could enable someone to match the file on "User 927" to a person. For example, if someone repeatedly does an "ego search" on his or her own name, that file might have the name in it repeatedly.
"It is not anonymous if the companies are tracking the same user over time," said Ari Schwartz of the Center for Democracy and Technology, an advocacy group that has filed comments with the FTC.
Especially troubling, he said, is that the targeters can monitor what people are reading, whether it's news or dinner recipes.
Underlying the FTC debate among public interest groups and Web media and advertising groups such as Google, eBay, newspapers and magazines are larger questions about the Web economy.
With surfers accustomed to accessing online entertainment for free, media companies have been pushed toward online advertising, rather than subscriptions or fees, to make money. But by many measures, online advertising revenue has proven disappointing.
While television advertising amounts to $64 billion annually, online advertising amounts to $11 billion, according to TNS Media Intelligence. Even at hugely popular sites, such as YouTube and social networks such as Facebook and MySpace, which each count tens of millions of visitors a month, owners have struggled to make money from ads. Television networks and newspapers, too, have seen that online advertising generates but a fraction of what they receive in print or broadcast, even on a per-person basis. Behavioral targeting promises to bolster sagging online ad revenue with a more profitable approach.
Most online ad targeting is relatively unsophisticated. Advertisers might know the geographic area of a user based on his or her Internet address. Or an advertiser might target a user based on the context of the Web page being read. An online magazine for audiophiles is a natural place for stereo ads, for example.
But if there is no obvious product to pitch on a Web page, the value of the ad space may be very low and is likely to attract only low-paying ads such as those flashing ads about looking for a lost classmate.
What behavioral targeting does is allow advertisers to target ads based not on what's on the page but who is looking at it.
Revenue Science, like other ad targeting services, tracks users by placing a "cookie," or small file, on a computer when it connects to one of the 3,000 Web sites that the company works with. The cookie essentially identifies that browser as a visitor to sites working with Revenue Science and gives each one a randomly assigned number. No names or other personally identifiable information, such as age or address, are recorded in the cookie.
When a user visits such a site, Revenue Science can record what pages were viewed, what search queries were entered and other information. It can even count, if a newspaper or other publisher allows, how many times a person sees a story regarding any given search term, whether it is "al Qaeda" or "denture adhesive."
The practice becomes more powerful as users move from site to site, betraying more information about their tastes. While some Web sites refuse to share their behavior files with other sites, the ad networks offer financial incentives to Web sites that do. (The Washington Post, which uses Revenue Science, does not allow Web behavior from The Post's site to be accessed by others.)
Detecting from previous Web visits and searches that a reader may be interested in new sport-utility vehicles, for example, a Web site can make as much as 10 times the amount of money showing an ad to that user, compared with an undifferentiated ad.
But while the tactic may lead to profits, it also creates unease. A March poll by Harris Interactive showed that six in 10 people are not comfortable when Web sites use information about a person's online activity to tailor advertisements or content.
As the long-brewing debate shifts to the Federal Trade Commission and possibly Congress, newspapers are likely to play a leading role. The Newspaper Association of America has filed a brief with the FTC arguing that some of the voluntary rules proposed by the agency's staff might violate the First Amendment.
More than 600 newspapers have formed an ad consortium with Yahoo. Another company, formed by the New York Times and three other chains, similarly offers advertisers behavioral targeting.
"The problem for newspapers is that a story headlined 'Two Dead in Baghdad' isn't very product-friendly," said Kent Ertugrul, chief executive of Phorm, a behavioral targeting company working with British newspapers. "But if you know who is looking at the page, that's where the opportunity is."
It is just such added revenue, newspaper lobbyists argue, that the troubled newspaper industry may need to survive the online transition.
In its first draft of voluntary guidelines, the FTC staff called for clear warnings of tracking and for allowing users to permanently opt out of a Web site's tracking mechanism.
"Every Web site," according to the FTC's draft rules, should allow consumers to "choose whether or not to have their information collected for such purpose."
But the newspaper association argues that allowing the user to opt out isn't necessary: If a user doesn't want to be tracked by a site -- assuming the user is aware of being tracked -- he or she can simply avoid that site. Besides, Boyle noted, users are free to periodically delete the cookies on their computers.
"I really don't know that there is a personal privacy issue here," Boyle said. "The government really needs to let things play out."