• by Steve Smith , Staff Writer @popeyesm, November 29, 2005

We're all used to patting down Google, Yahoo!, or MSN for relevant information, but what if they started frisking us first? What if the search engines we've all come to rely on started analyzing our online activity, or frequently searched terms, and used that information to determine, let's say, how close we were to making a purchase, and which brands we were considering? Would marketers pay for such information? If so, how much? Would consumers accept this degree of targeting, or would they resist it as an invasion of privacy? And will search marketing, which has worked so simply in the past, be able to absorb this new level of complexity?

Advertisers already pay to match their automotive ads with searchers entering "cars," but what if the engine also found out which searchers were high-end buyers more likely to respond to a pitch for a Lexus RX than a Chevy Cobalt?

The prospect raises a host of questions: How much more would marketers pay for this level of intelligence? How would search engines provide such information? And, assuming technology can glean this level of detail, will consumers acquiesce?

We may know some of those answers sooner than you'd think.

MSN's adCenter already has the ability to target contextual ads by demographics, geography, and day parts. By next spring, msn plans to add behavioral segments to the mix.

Meanwhile, revisions in Google's privacy policy, along with a patent application for "results-based personalization of advertisements in a search engine," suggest that it too is considering incorporating user profiles and activity history into content and ad targeting. AlmondNet recently received its second patent for delivering behaviorally targeted ads via its own engine. And adware provider Claria put into alpha testing its Vista Marketing Services platform, which personalizes search results and ad delivery according to a user's previous post-search behavior. Yahoo! also has a product, Yahoo! Impulse, that allows targeting of its display ads within its content network based on a user's recent searches.

MSN's plan to fold behavioral targeting into the wildly successful contextual text ad marketplace raises the bar even higher on targeting and pricing.

Microsoft hasn't yet outlined the specifics of the plan, other than to state that the company "will provide advertisers with the audience intelligence to help them more precisely identify their target customers and manage their campaigns."

MSN is promising unprecedented "transparency," giving ad buyers more direct views of actual MSN audience data in order to make ad placement decisions based on several targeting considerations that are new to search.

AdCenter's behavioral piece will join other targeting options, including geography and day part. The engine can apply several filters, so that the choice of ads on a result page could become highly individualized. AdCenter is premised on providing marketers with that level of individualized results, says Forrester Research analyst Charlene Li. "If they know I am in the market for an SUV or a hybrid, then ideally those ads should show up," she says.

The question for marketers becomes, how much more will they pay to know the person they are advertising to visited, say, the SUV pages of MSN Auto before typing "cars" into the search box? Rather than buying ads against words and context, you are buying against people and their likely habits. "What adCenter is doing already is changing the game," says Li. "You can already target day parts and geography, and now add to that behavior."

Media buyers are likely to pay a percentage premium on top of their keyword price to get these additional layers of targeting. Behavioral targeting itself, long on the periphery of online advertising, could move center stage with the entry of majors like MSN and Google, says David Berkowitz, director of marketing, Unicast. "It absolutely legitimizes BT. It becomes one of those things that is part of an interactive buy. It forces education."

Increasing Complexity

Some education will be necessary. Behavioral targeting already suffers from a fuzzy definition and myriad competing methodologies. Adding it into the search mix could initiate yet another round of feature wars and rising complexity among the major engines as each tries to differentiate its method of accurately determining behavior, as well as helping buyers manage more complex campaigns.

While details of msn's approach are scant, it appears that adCenter will profile user behavior based on recent content browsing patterns within the portal's collection of sites, and use evidence like multiple searches on the same term to indicate behaviors such as the level of purchase intent.

Google, which does not have MSN or Yahoo!'s content networks or their volume of registered, profiled users, may rely more on harvesting profiles from the myriad client-side applications (desktop search, Gmail, toolbar, etc.) it has been pouring into the market in the past year. In its new patent application, Google indicates that its profiling mechanisms could be embodied in desktop programs.

BT-enabled search introduces new methodological debates into the straightforward search landscape. For instance, adware company Claria hopes to sell engines and portals on the superiority of its own client-side solution: an always-on toolbar or desktop tool that assembles richer profiles from the full range of user behaviors rather than activity occurring within a site or portal. "If you only rely on site-specific information, you don't see over 95 percent of people's interests," says Scott Eagle, Claria's chief marketing officer.

This type of dense profiling could ultimately distinguish a high-end business traveler from a student hostel-dweller before she puts "China travel" into that search box. A BT-enabled engine would know whether to serve her an ad for the Beijing Hilton or CheapTickets.com. "We believe the next quantum leap is understanding what people do and how they act when they search," Eagle says.

Rather than build a new search engine of its own, Claria is actively looking for big partners that could take its software to sufficient scale. In fact, Microsoft reportedly looked to acquire Claria several months ago, but ended discussions because of possible pr backlash from buying an adware firm. Eagle would not comment on Microsoft's interest and whether it included Claria's search solution, but the thwarted courtship is yet another sign that incorporating BT into search comes at some risk to trusted big brands.

Promising? Yes. But there is enough confusion surrounding BT's techniques, execution, and privacy implications to make it a dicey addition to the simple, familiar search ad buying process. "It's going to open up a slew of new questions," Berkowitz says.

Indeed, one of search's major advantages has been its sheer simplicity. Adding behavioral targeting elements into search marketing is far from simple. "It removes the kiss [Keep It Simple, Stupid] element which has worked so well in search," Berkowitz says.

Consumer Backlash?

In the first half of this year, search revenues soared to $2.3 billion, marking a 27 percent increase from the same time last year, according to the Interactive Advertising Bureau. At the same time, users have steadily increased their visits to search engines. In August alone, visitors conducted more than five billion searches, according to Nielsen/NetRatings.

With increasing search engine involvement in behavioral targeting, can consumers be expected to revolt? Even if the data is not "personally identifiable" and marketers don't know consumers' offline information, some are bound to find such ads intrusive. A simple wording change to a Google privacy policy last month prompted scores of news and blog stories warning about the engine's intentions.

Some industry insiders have said they consider BT somewhat "creepy," and are concerned that its techniques aren't always visible to the marketer, that its monitoring practices may seem invasive, and that it can appear to stalk users with ads no matter where they go. Unlike buying against words or pages, where the placement is obvious to any client who puts in his keyword, BT has a lot of moving parts behind the scenes that serve ads invisibly to select users.

Ultimately, issues like privacy and profile accuracy may turn on how well the engines manage relationships with users and create a fair exchange of value. Will users agree to have their behaviors tracked, even anonymously, if the only payoff is a more targeted text ad in msn? Or can engines obtain even deeper, more accurate, and necessarily more invasive profiles if the search results themselves become more personalized, or if the data harvesting occurs on valuable tools like desktop search, or even a full-blown application?

For instance, a partnership with Sun Microsystems allows Google to help distribute OpenOffice, a free, open-source alternative to Microsoft's pricey Office suite. Along with its toolbar, desktop search, Gmail, and a relentless line of other free tools, Google is amassing a line of powerful freebies it could turn into behavior monitors if users agree to exchange their data for free programs. "It all depends on the mission of the company and how they approach their customers," says iCrossing's Noah Elkin. "I think it's a model consumers are becoming accustomed to."

But will other publishers become accustomed to such models? Search engines are platforms, virtual utilities that are interdependent with all other online content. Incorporating BT into such ubiquitous services will likely raise questions from publishers about where and how engines do their harvesting. For instance, if a major search engine's desktop client monitors user behavior on third-party Web sites, it would be taking something valuable to a publisher: its user data.

As it stands, BT companies like Revenue Science and Tacoda financially incentivize publishing partners that come into their networks and share anonymous behavior data. Adding BT to search is a smart thing to do, says Bill Gossman, CEO, Revenue Science, but "using a client side application to do that would be grabbing the third rail of the industry and standing in a bucket of water."

And just as all aspects of search engine methodology are hotly debated, there are bound to be questions about the nature, breadth, timeliness, and accuracy of the profiles that inform the segment that marketers buy. Millions of users may start Microsoft Hotmail accounts or sign up for Google Desktop search using some false or misleading information, which could affect how they are used for targeting. "Behavioral targeting is as good and as accurate as the data itself," Elkin warns.

Profiling is more an art than a science. When a vendor offers up a predefined segment such as in-market HDTV buyers or self-directed investors, that profile is based on a set of assumptions made only on the basis of click behavior, not direct questionnaires. In other words, profiling on the basis of browsing behavior is an interpretation -- and like page ranks and content matching algorithms, the procedure will differ from vendor to vendor.

With challenges and complexities this profound, some observers think that Google and Yahoo! might hold off on incorporating behavioral targeting in search until they see whether msn suffers any user backlash. "This could be one of those instances of a first mover disadvantage," says Unicast's Berkowitz. "[MSN] might monetize it like never before, but there is more than one possible outcome here."