Keeping BT in Context
As curiosity and excitement about behavioral targeting continue to build, the natural temptation among publishers and networks is to sell it (and for advertisers to avidly buy it) at a high premium. This temptation, general manager of ValueClick Dave Yovanno explains below, needs to be tempered by reality. The value of BT, he reminds us, needs to be judged in the context of how it relates to overall Web site or network optimization.
Behavioral Insider: What do you see as the biggest misconception advertisers and networks have about behavioral targeting?
Dave Yovanno: A big misconception, and it's aided by publishers and ad networks for obvious reasons, is that it makes sense to charge and pay a big premium for behavioral segmentation because it will provide an enormous lift in ROI by itself.
BI: You think there's a danger BT is being oversold?
Yovanno: Don't get me wrong. Behavioral campaigns can and do score great performance improvements, but these need to be seen in context. In order to get really sustainable results they need to be based on a really significant amount of relevant data -- which doesn't just grow on trees. It can't just be started from scratch by buying a lot of ad inventory and then waiting for a large enough segment to develop to be able to really scale a campaign targeted at fuel-efficient vehicles, say.
To really make systematic use of BT, you need a critical mass of unique users and a good deal of relevant data from all four aspects of behavior to leverage. So behavior's a powerful piece, but becomes much more so in the context of overall optimization of all sorts of data. Methodology and technology without scale is of very limited usefulness. Of course the opposite is true, too.
BI: Where do you see behavioral targeting as fitting into what ValueClick is doing overall?
Yovanno: We didn't get into behavioral targeting from the perspective of showing it as a standalone kind of feature. Our interest in using behavioral data stems strictly from our interest in finding ways to get the best results for our campaigns. ValueClick's forays into behavioral have been based on understanding the maximum amount of a wide gamut of attributes that relate to better predicting performance. In our world it's an extension of the kinds of optimization we've always tried to do. Behavioral targeting is an important layer, a value-added layer atop an existing network.
BI: What types of behavioral data do you work with?
Yovanno: There are four types of behavior we can track: browsing sites, search, purchasing and ad response. One of the most important aspects of behavioral targeting is to learn how to use each of the four main categories of behavioral data. It's important to try to relate each piece, but one of the most important things is to understand and learn from how users respond to different types of creative and messages.
BI: Re-targeting is one of the most popular areas of BT at this point and something I know you introduced in your network. How do you see that relating to the wider realm of behavioral targeting?
Yovanno:Re-targeting is certainly the starting point for behavioral and where it's gotten the most traction, but the bigger issue is whether or not you're consistently learning how to target better with every ad you serve.... It's a matter of tweaking creative and format certainly, but also of locating where and when receptivity to ads is strongest and where they are not being responded to. But also of identifying where customers are in the sales cycle, which will be different in each vertical.
The opportunities are tremendous if you can pull all the behaviors together. For instance, a great example is in auto, where we're looking at a period of ten weeks when consumers are in-market. Only about 5% of that time on the Web is spent directly at auto sites. The rest is spent surfing to other sites.
BI: Can you discuss the differences between how direct response and brand advertisers currently see BT?
Yovanno: Our direct response advertisers expect results. They still don't know or care about whether they're due to behavioral targeting or not. It's strictly a means to an end. The first thing brand advertisers think about is which sites they're going to be associated with. It's harder - or, let's say, not as obvious -- for them to think of reaching an audience outside a particular media site, so getting traction in that space has taken longer.
BI: What are the most important learning challenges ahead for advertisers in better utilizing BT?
Yovanno: We look at over 100 attributes of ad and site performance and are trying always to add more. No data piece is necessarily a panacea in itself. All have a more or less important place in certain circumstances. The more real-life user data you can work with, the more meaningfully you can understand what your metrics mean, and what sorts of time intervals between ad impressions and clicks, and between clicks and conversions, are expected and acceptable.