As a media buyer, you really have to dive into the weeds to find out how ads are being targeted. But understanding BT methodology pays handsome dividends.

It's not uncommon to see 1,000 percent lifts in clickthrough and conversion rates when adding behavioral targeting to an online ad effort. With that kind of performance differential between targeted and untargeted ads, it's more important than ever to understand how your ads are being targeted.

Behavioral targeting has captivated marketers and agency people alike with promises of increased return, but it wasn't until fairly recently that behavioral targeting learned how to scale appropriately. For example, when Tacoda was in the enterprise software business, its clients realized that BT's biggest issue wasn't identifying people according to their web behaviors but rather how to find enough people fitting certain behavioral criteria to make campaigns worthwhile.

When Tacoda changed its business model and moved to establish a network, it began to overcome some of those scale problems. When it was acquired by AOL, many hoped that AOL could bring enough ad inventory to the table to really help Tacoda realize its full potential, but I've already covered that subject.

But even when we hook up a great targeting technology to a huge ad network or portal, scale problems still don't go away entirely. This is why BT is best represented as a series of concentric circles on a PowerPoint slide -- the bull's-eye represents the universe of people who fit your behavioral criteria exactly. Working outward from there, the rings of less-desirable prospects fall into those rings because they fall into a group where either the criteria are relaxed somewhat, or they're matched to the ideal prospect through the use of a surrogate.

For example, you might define your audience as leisure travelers. Behaviorally, you might have established that they visited leisure travel content at least twice in the last 30 days. There are sites and ad networks that can deliver impressions against that behavioral profile in its current form. The first concentric circle outside the bull's-eye, though, might be a bucket of users that meet the site or network's definition of "traveler." It's still a desirable audience, and you'll likely want to reach them, but they may not be as close to your target as you might like. For instance, a travel bucket might also have a large number of business travelers in it who don't meet your target definition.

Another concentric circle might involve "look-alike targeting," which essentially takes an audience and locates more people that look like the original target audience definition, when expressed through web behavior.

By way of example, this is how Yahoo targets people within its audience in programs like Consumer Direct. It zeros in on a group of people based on their offline shopping behavior (via a partnership with Nielsen), and then it scores them against their visitation to several hundred content categories. Yahoo refers to this scoring as "behavioral thumbprinting" and uses that thumbprint to find people within its audience that match the type of person who displays the desired offline behavior.

As you can see, this method uses behavioral targeting as something of a surrogate. It would be impossible to find a large number of people who fit the desired behavioral criteria, even within Yahoo's vast user database -- so BT is used to project to Yahoo's user base.

Sometimes as a media buyer, you really have to dive into the weeds to find out how ads are really being targeted. But understanding BT methodology pays handsome dividends -- you get an understanding of just how desirable various behaviorally targeted inventory can be, and you'll get better at predicting performance as well.

Over the long haul, you should develop a hierarchy of desirable BT types and how compatible they are with your business goals.

Tom Hespos is the president of Underscore Marketing  and blogs at Hespos.com.