Targeting may be a numbers game, but it isn't a science, yet. See what ValueClick SVP Matthew Boyd thinks the future will hold for BT.

There's a lot of data out there, but most of it is living in the past, according to Matthew Boyd, SVP at ValueClick Media.

The trouble with behavioral targeting as we've come to know it, Boyd said, is that it's only good at determining what a user will likely do based on past data. That's been a real breakthrough for online advertising in the past few years, but Boyd sees a new trend emerging -- predictive BT.

Speaking at the iMedia Agency Summit in Austin, Texas, Boyd made the case for what he called the predictive approach to BT, a technology ValueClick is in the process of rolling out. According to Boyd, the shift has been brought about because online advertisers have become less concerned about pages and more interested in people and audiences.

"There's this shift underway from targeting pages to targeting people," Boyd said. "[Targeting] pages in a contextual way works extremely well if you are looking to get a consumer based on the person's current behavior. But there's a lot of waste, too. Some people just drive by, but you also miss the enthusiasts potentially who are going through tons of pages."

According to Boyd, that approach does an efficient job of putting people in buckets, but those buckets block marketers from determining what consumers' future interests are likely to be.

"Put simply: All forms of BT simply assign folks based on past behavior," Boyd explained. "Predictive isn't what you've done in the past, it's what you're likely to do in the future. With predictive you try to overcome the limiting factor of the cluster approach because people fall into so many different categories."

But where the old model required several data points (how many visits a user made within a given time frame, what kind of content the person searched for, etc.), the predictive approach demands an even greater amount of intelligence. So where does it come from?

In a nutshell, Boyd sees predictive BT as stemming from four places on the web. By leveraging the company's four distinct businesses (affiliate marketing, ad serving technology, comparison shopping and display advertising), ValueClick is betting that it knows a great deal more about what people are likely to do in the future than do most advertisers, or even consumers.

"Predictive BT dynamically assigns visitors to segments based on recency, frequency and likelihood of future performance," Boyd said.

According to Boyd, predictive BT is a waste-saving tool that helps marketers discard so-called "drive-by visitors" and enthusiasts who may rack up a high page count of relevant content without really being in the market for a given product.

Consider the example of an automotive campaign. According to Boyd, the value of an automotive enthusiast and a random clicker is about the same -- zero. Why? Because neither is truly in the market for a car, so why waste ad dollars targeting them?

The challenge, Boyd said, is to find the buyers who are in the market for a car, even if they may not be signaling their intent by consuming page after page of automotive content. For ValueClick, finding those people will require the ability to harness nearly a decade of data from multiple business groups. 

Michael Estrin is associate editor at iMediaConnection.