• by Phil Leggiere , October 8, 2008

The core value proposition of behavioral targeting is to personalize advertising impressions in an unprecedented way by focusing on "users rather than pages." In practice, however, Dominic Bennett, vice president of engineering at Turn.com, explains below what's been missing is a pricing model flexible enough to link specific behaviors and behavioral segments to specific goals and strategies.

 

Behavioral Insider: Could you explain the motivation behind your introduction of the dynamic pricing model for behavioral targeting?

Dominic Bennett: Our motivation for getting involved in behavioral targeting goes back to our core proposition as an ad network. From our inception, we've looked to find different ways to introduce new ways to optimize ROI. These have involved improving options for automating targeting and prediction.

So what we tried to do was put ourselves in the shoes of advertisers, as it were. Advertisers see behavioral targeting as a means to an end -- and that end can and does vary greatly, not only from advertiser to advertiser but from campaign to campaign. Depending on a particular circumstance, you may be trying to drive sales, to raise awareness, to generate leads, and to expand engagement with existing customers. Often an advertiser will have multiple goals.

BI: What were the major problems with conventional behavioral pricing models? How have the value and ROI propositions of behavioral changed as marketers have become more experienced?

Bennett: The biggest limitation of the way behavioral targeting is offered has been that pricing has been based on a one-size-fits-all model. The idea has been that behavioral targeting is a premium service, and you need to buy it at a fixed CPM rate.

A very simple basic example is, if our data indicate that a Web visitor, based on their behavior, is high in a funnel, an advertiser will quite likely be prepared to pay more than they would for a visitor who's at a lower level of the funnel. At the same time, strategically they may well want to reach people at a variety of levels within the sales funnel. Why shouldn't they be able to price impressions depending on the specific campaign needs they have, and their ROI requirements and goals? Behavioral targeting is very complex, and if you have a very granular system you can identify a wide continuum of data points and yield a very detailed range of segments. So when you think of it, it's kind of absurd to assume, as current models do, that this variety can be captured in a fixed CPM.

BI: What are the advantages of having a multi-tiered flexible pricing model for deploying behavioral campaigns?

Bennett: What we see this doing [is] challenging the fixed-tier model. This has strong potential upside, both for advertisers and publishers. For publishers, it's a new way of looking at their inventory and its potential yield. If behavioral targeting in its early days taught us to think in terms of users as well as just pages, the new approach teaches us that not all users or behaviors are created equal. Their value at any time is in the eye of the beholder.

BI: How well-positioned is behavioral targeting for what appears to be a possible recessionary trend in online display spending?

Bennett: There's clear-cut evidence that if you deploy behavioral data correctly, integrating multiple data points, you get huge improvements of results versus run-of-network, certainly. But we also have a new prevalence of what I call lookalike profiling. One of the big problems right now -- and it's only going to get more pronounced in a slow economy -- is that there are so many providers offering behavioral targeting.

What I mean by that are attempts to jack up the scale by grouping together a strong behavioral profile that shows someone is, for example, an in-market auto shopper, with far weaker profiles based on people with similar behavioral patterns that 'might' indicate certain interests. This sort of thing dilutes the impact of behavioral campaigns.

BI: What are the biggest challenges and opportunities of the space going forward into 2009?

Bennett: Looking forward, one thing we're quite focused on is developing metrics that measure long-term value. Another important criterion that deserves a deeper look is engagement. We are trying to factor in recency as a barometer of engagement as well as concept of velocity to measure whether, for instance, a consumer is becoming more or less engaged relative to where they were.

We really feel that the effect of a recession will be that advertisers will become, of necessity, much more motivated to separate the wheat from the chaff. One thing's for sure, the days of overpaying for questionable performance are over.