Striking a Balance with Ad Targeting

• by Tom Hespos , June 7, 2005

Online advertising has more targeting options than it has ever had, and behavioral and contextual marketing are two of the hottest concepts in the space. While these concepts are not new to most interactive agency planners, they are relatively new to many clients. And as with anything new, many clients are still in a period of adjustment regarding targeting and how they'll use it to achieve online objectives.

This afternoon at the OMMA West conference in San Francisco, I'll be moderating a panel on the publishing track that deals with how clients and agencies are using data overlays to target their ads. One issue we'll be dealing with is client adoption of behavioral targeting and data overlays in general.

As advertisers circle around the notion of targeting, dipping their toes in the water with tests, I see a few conceptual difficulties that often arise. The first is the notion of targeting based on a displayed behavior when clients tend to be used to a contextual approach. Running a behaviorally targeted campaign often takes advertising out of the context it has traditionally occupied. In a behavioral campaign, the inventory is often taken from Run of Site (ROS) or areas that may not make sense contextually in pursuing prospects meeting a set of behavioral criteria, and this is often difficult for clients to grasp after they've been running ads in context.

It's a good tradeoff. Behavioral campaigns can often show a lift of 30X or better when compared to ROS or RON campaigns. But when clients are used to taking an approach that involves sponsorship of relevant content, letting go of context in favor of desired behavior is not without its conceptual barriers. What we need to do as an industry is be sure to measure behavioral inventory, and contrast its effectiveness with what we've been doing in the past. And we need to do this in terms of easy-to-understand numbers. It will help clients understand when they need to let go of context and place their faith with behavioral targeting algorithms.

But even when clients get their first taste of behavioral targeting, they often need to be coached with regard to their expectations. The tendency is for clients to use data overlays to overtarget their ads, concentrating on the "sweet spot" of their media objective and using targeting filters to concentrate too much of their ad buy on a small universe of people. We did some work for a client last year who wanted to spend tens of thousands of dollars targeting people who had purchased at four e-commerce sites. While this universe certainly did represent the sweet spot of their target, it consisted of only a handful of people who ended up seeing the ad so many times that they became sick of it. Despite protests from the agency and strong suggestions that they loosen the screws a bit when it came to their targeting parameters, the client pressed forward--and suffered because of it, ignoring hundreds of thousands of people that would have been interested in their product.

Granted, there are quite a few advertisers that have tested data overlays, and have struck the proper balance with them. But we need to adjust expectations for clients who are just getting on board with complex ad targeting.

To do that, we need to get clients to understand two critical axioms of marketing in the behavioral age:

1) Behavior is a strong predictor of interest in a given product or service, and in many cases it is a better predictor than context. Letting go of context in favor of behavior can often be a tactic that produces strong results.

2) Targeting can be abused, particularly when it limits an audience universe more than it needs to.

To get clients to better understand these concepts, we need to express their buys in terms of the audience they will reach. An approximation of reach and average frequency against an audience, as well as an estimation of the size of the universe given the behavioral parameters, will help.