Monday, June 04, 2007

The Right Kind of Behavioral Targeting

The recent acquisitions of DoubleClick by Google and RightMedia by Yahoo are rightfully raising privacy concerns about behavioral advertising. E-Commerce Times has reported that the FTC is investigating these deals because "Privacy issues and antitrust concerns go hand-in-hand, since the mergers mean combining overlapping databases about consumers' online behavior, from search terms and e-mails to digital photo collections and advertising clicks." And EU regulators are concerned as well, particularly due to the length of time Google stores behavioral information on users.

Concern centers around cookies. While each cookie placed on your computer is fairly limited in what it can reveal about you, the entire collection reveals quite a bit. And with just a few companies now having access to the information in all of those cookies, your privacy is significantly diminished. Consumer privacy groups worry that "
Google’s proposed acquisition of DoubleClick will give one company access to more information about the Internet activities of consumers than any other company in the world." Think Tom Cruise in Minority Report.

The notion is that if an advertiser knows the last 100 things you've done online, he or she can place targeted ads in front of you that you have a high likelihood of clicking. But what if that theory is just plain wrong?

Jack Jia, founder of Baynote, has a different idea—a concept he believes will work better than behavioral targeting and be far less invasive to individual privacy. Baynote uses group behavior to improve site search and advertising. According to Mr. Jia, "Past behavior (i.e. tracking someone's every online movement) is a very poor predictor, because humans have way too many profiles. I am a father, a son, a brother, I like travel, I like a lot of things. You can track all my past behaviors all you want, but in any given moment when I go onto a site it is very hard for you to predict what I want...We are raised with the notion that I am unique and don't have needs quite like other people. But the scientists proved we are not unique at all. We are pack animals. Pretty much 95% of people will need the same thing. We only need to find out under what context what things are useful, and then present that product or content given the context. Then the prediction is very accurate."

Baynote's approach has also been covered on The Next Net blog and Software Abstractions, and according to MarketingRev, "Baynote stood out (at the recent Web 2.0 conference in San Francisco) because they do something different and highly compelling. They continuously analyze traffic patterns on a Web site’s pages in order to dynamically optimize search engine results, navigation and content. In essence, they leverage the actual behavior of the crowd to dynamically shape what gets served up to subsequent visitors."

This notion also fits well with a couple of observations made by author Nassim Nicholas Taleb in The Black Swan: The Impact of the Highly Improbable (reviewed here). First, that the ultimate success of particular movies, actors, books and other "products" are driven by their initial success due to the imitation factor: "Such contagions (e.g. the viral effects of a big opening weekend for a movie) do not just apply to movies: they seem to affect a wide range of...products. It is hard for us to accept that people do not fall in love with works of art only for their own sake, but also in order to feel that they belong to a community. By imitating, we get closer to others—that is, other imitators. It fights solitude."

Second, that group membership is a good predictor of behavior—as long as one looks at the right groups. The groups that we choose to be a part of (e.g. profession, hobbies, club and association memberships, etc.) are strong predictors of behavior; the groups that we are naturally a part of (e.g. skin color, height/weight, ethnic origin, etc.) are poor predictors (with the general exception of gender). For example, an upper middle class Irish-American medical doctor will have far more in common with an upper middle class Italian-American medical doctor than he will with a middle class Irish-American plumber.

What all of this means is that if you something about the voluntary group associations of a specific person, and you know how a particular group tends to act online, then you can predict behavior (and target ads) with a fairly high degree of accuracy. Collecting any invasive individual information beyond that is not only an assault on privacy, it's counterproductive—it won't make predictions any more accurate, and may very well make them less so.

All of which means the future of online advertising doesn't have to resemble the world of Damon Knight. It would actually be much better for both individual privacy and marketing effectiveness if it didn't.


Minneapolis marketing and PR agency exclusively focused on B2B technology companies: KC Associates

The Internet marketing promotion portal:

Contact Tom Pick:


KudoSurf Me!
Add to Technorati Favorites
AddThis Social Bookmark Button
AddThis Feed Button

eXTReMe Tracker