• by Robert Tas , November 22, 2004

By Robert Tas

By now everyone's heard of behavioral targeting. You know, the hottest thing to hit the online marketing space since paid search? Simultaneously, it's the publisher's miracle cure and the advertiser's secret weapon. The noise surrounding behavioral targeting is deafening with too many definitions, too much revisionist history from "targeting" providers and too much hype.

It's time to stand back, reflect and move toward a system that defines behavioral targeting properly, deploys it in a logical fashion, and leverages it for what it is supposed to be: a tool to employ audience data to increase campaign performance for advertisers and publishers to extend highly valuable inventory into lower demand areas.

At the end of the day, behavioral targeting is just targeting, like dayparting or geo-targeting. That said, there are a handful of key definitions and precepts that must be considered when thinking about what behavioral targeting is, how it can be useful to publishers and advertisers and what results we should expect from it.

Definition Let's first dismiss what behavioral targeting is not: Behavioral targeting does not consist of solely watching a consumer's interaction with a banner, or other ad unit. That is called optimization and almost every basic ad server does this in one manner or another.

Behavioral targeting is aggregating observed consumer behavior, segmenting the audience via a rigid and robust taxonomy, and delivering the right ad to the right person based on segment data. To put it another way, behavioral targeting is understanding how the frequency of a user's interaction with a brand over time reflects that user's mindset and purchase intent.

Behavioral data is rich and multi-dimensional, since it is built from the bottom up on audience behavior. It takes into account the content category visited or products purchased, how frequently these occur and then layers ad interaction on top of it.

Companies need to begin looking at a data strategy and understand that starting small is okay. Audience segments will evolve through the interaction between the advertiser/agency and publisher. There will be obvious segments and obscure ones but the key is being committed to moving toward selling audience, and not pages. We need to listen to people such as David Verklin, CEO of Carat Interactive, who asks us to speak the same language as traditional media players.

Scale This is somewhat qualitative, but audience size must be a determinant in leveraging behavioral targeting. All advertisers and publishers want to hit the sweet spot of prospects, but we must avoid the tendency to create unrealistically precise targeting that creates a segment so small that it is not viable for the advertiser or the publisher.

I put segments to a commercial viability test by asking: Is the size of the segment large enough to warrant the effort needed to manage it, create the messaging for it, target ads to it and buy inventory in it? Be wary when behavioral targeting companies seek to employ raw clickstream behavior to construct segments. This is a telltale sign that the targeted segments are too small to matter.

Publishers should roll out segments only when they believe they are big enough to service, sell, target, and bill. To be salable over time, segments also need to be reproducible throughout the publishers' sales force, just like other ad products. The sales organization must be committed to the product, or it will fail.

More to come next week on the performance, integration, and segmentation of behavioral targeting.