7 Facts About On-Site Behavioral Targeting

In advance of my speaking on Tuesday July 24, 2007 at MediaPost’s Behavioral Targeting Forum in New York City along with OTTO Digital’s Chief Creative Officer Paco Vinoly, I thought I would share some of the things we’ve learned to be true in doing Behavioral Targeting (BT).

Fact 1: Behavioral Targeting isn’t new, nor is it difficult to execute.

Any of you that market online for a profession will at some point within the next 12-18 months start doing BT onsite for copy, products, promotions, video, messaging, even certain widgets and applications. If you are a search marketer you have already been doing this for years and maybe didn’t even realize you can put it on your resume.

Fact 2: Segmentation is the key to effective targeting.

Targeting can be accomplished through Source, Behavior, Temporal & Environmental segments. Creating these segments and discovering where the highest impact can be made is a major key to success.

Fact 3: Simple targeting yields the best results.

Creating high-impact segments has multiple benefits. Foremost these segments are easy to define. They can be as simple as source segments of Google vs. Yahoo or behavior segments such as 1st time visitor vs. return, temporal segments like weekday/weekend and environmental segments like Geo or Screen resolution. They also have the largest populations. This is important so that we achieve confidence in the metrics by reducing margin for error with a larger data set. It is also where you can really move the results needle. Even small lifts in performance here can have great overall impact.

Fact 4: The further down you get in the user goal funnel the easier it becomes to deliver a more relevant experience.

No offense to ad targeting. I think there is a real place with this at the top of food chain. However, once a user has already displayed intent “the conversation” has started. An ongoing conversation with a visitor makes it easy to define relevance. That conversation may also span multiple visits or transactions.

Fact 5: User behavior can be segmented into groups of similar actions.

Discover patterns of behavior. Self-segmentation is a useful strategy where it can be employed. The affinity algorithms work very well with this as a factor. In many respects every click is a form of segmentation. You may already be sitting on a goldmine of segment data in your analytics.

Fact 6: For best results, segments once created need to be validated.

In testing so much of what we see is counter intuitive. Marketers should never assume that they know what customers will respond to, even at the segment level. Segments should always be validated with testing against a control group of more inclusive traffic.

Fact 7: BT requires more resources for creative.

No longer are we creating a single version for everyone. In many cases segmentation and targeting require 3-4x the creative work. Make sure you have alloted the time necessary for this in your project plans and pricing.