• by Tom Hespos , June 17, 2003

We've always known that what consumers do and what they say they do are often two completely different things. That's why direct observation of a consumer's consumption habits tends to be more accurate than using recall-based surveys. That's why it's tough for me to fathom why observed behavior of online consumers doesn't play a larger role in the targeting of online advertising.

Not that it hasn't been tried before. Engage once had an entire ad network called AudienceNet that was based on anonymous behavioral targeting. Sites like Yahoo have had behavioral offerings for a while. There are a few desktop applications that track surfing behavior and can target ads in that fashion, as well. Yesterday, an article in The New York Times highlighted behavioral targeting offerings in The Wall Street Journal Online and other websites. It's clear that targeting by observed behavior is quickly becoming a trend.

But why? You don't hear much about Engage's AudienceNet any more (unless you're taking some of your interactive media buyer friends for a quick stroll down Memory Lane). But you are hearing quite a bit about how behaviorally-targeted advertising can produce results orders of magnitude greater than untargeted online advertising. The difference between early offerings like AudienceNet and modern behavioral targeting offerings is in the execution.

Three things are necessary in order to have success with this model:

• Dynamic audience targeting capabilities - In essence, media buyers like to define their own audiences. Creating static audience buckets will increase the value of online advertising in certain categories, but allowing media planners to create their own behavioral profiles to speak to the kinds of people they like to speak to is key.

• Scalability - Dynamic audience targeting capabilities are great, but the thinner a media planners slices an audience, the smaller the universe. Publishers need to be sure they can scale small audiences by making sure that consumers who display a certain type of behavior can be reached not only in context, but elsewhere on a site or through other channels (e-mail, etc.). Otherwise, the behavioral model doesn't scale. Desktop applications have the advantage of being able to reach people any time they're on the web, which addresses the scalability problem. Website publishers can address it by retaining their audience, generating repeat visitation, and using technology to be able to target desirable people in out-of-context areas on their site.

• A change in thinking - Media buyers are going to have to become hip to the idea of targeting folks by behavior, rather than by demographics or context. For some media buying organizations, this is a no-brainer. For others, it will take some coaching. We have to give planners and buyers case studies and success stories so that they can fully understand the behavioral targeting concept, and get them used to targeting ads in alignment with their planning target, not their buying target. (e.g. - Outdoor sports enthusiasts, not Men 18-49.)

Now that behavioral targeting has some traction and has produced some compelling results, let's move this forward as an industry initiative and show marketers what we can do with this. We've learned a lot since Engage debuted AudienceNet. Let's show it.