• by Phil Leggiere , June 16, 2006

The topic of behavioral targeting has yielded no shortage of speculation and analysis, much of it, for better or worse, of the "40 thousand feet-high," long-view variety. There are signs, however, that a new stage of discourse, this one a little closer to the ground, may be upon us, as advertisers, networks and publishers begin to share their data and experiences with targeting. Dakota Sullivan, CMO of ad network Blue Lithium, discusses below an initial case study his network is preparing and goals for further documentation of client experiences.

Behavioral Insider: You've recently come out with the first round of case studies on client results from using retargeting behavioral methodology on your network. What do you see as the significance of this move?

Dakota Sullivan: We've claimed for a long time that behavioral targeting could give clients unique advantages over run-of-network--and had all sorts of anecdotal evidence supporting that--but now we're finally at the point where clients have a comfort level enough to share data points. That's a big step forward, not only in documenting what targeting can accomplish, but also [in providing] advertisers with a fuller body of knowledge to guide their own learning curve. It helps that the results of our first phase of case documentation were pretty dramatic. Three major clients shared their data with us, and the interesting thing we found was that targeting customers who'd previously visited an advertiser Web site but left the site prematurely yielded a 300 percent increase in click- throughs versus run-of-network, with actual conversion rates substantially higher than that.

BI: What was the scale of the campaigns?

SULLIVAN: These were big multi-month campaigns involving up to 100 million impressions. One of the customers, a major wireless phone company, saw a 152 percent improvement in click-throughs and 758 percent improvement with 100 million impressions. A large satellite TV company had a 232 percent click-through boost and lift of over 3000 percent in conversions, while a major online university saw a 299 percent jump in click-throughs and 337 percent more conversions.  We expect to announce their names publicly soon.

BI: How was this particular cluster of studies structured?

SULLIVAN: In this initial documentation we tried to keep it as basic and pure--if you will--as we could, targeting versus non-targeting. This data was developed under sophisticated testing conditions.  Each of these clients had a control campaign running under the same conditions, with the same creative, over the same period of time.  The only difference was in the targeting. 

BI: Beyond documenting that targeting outperforms non-targeting, were any other insights yielded? Any surprises you hadn't considered before?

SULLIVAN: We were surprised, certainly, by how sharp the increase was, but in addition, it was interesting to notice the differential between the results retargeting scored and the standard behavioral methodology of audience profiling, which has been positive, but not nearly to the same extent as retargeting.

BI: Do you see this group of studies as a first step? And if so, where do you want to see the next phase of documentation go?

SULLIVAN: In the real world, of course, the challenges of every promotional campaign are different--and so, once advertisers have moved into targeting, the next phase in documenting case study results will be to weigh more of the relative strengths of different kinds of targeting not only as they perform versus run of network but versus other kinds of targeting options, be it geo-targeting, demographic or day part, etc. Over the next few months we expect to get a lot more documentation on this. Beyond that you get into distinguishing results between using one behavioral methodology in isolation versus combining targeting methods. Of course the conundrum, once you begin drilling down into deeper levels of granularity, is finding what scale will provide you real intelligence. The more granular the targeting, the smaller the numbers.  Right now we encourage clients to begin by just testing to scale first--and then to layer their targeting.

BI: This week you launched a video network. Do you see the experiences reflected in the study as relevant to this initiative?

SULLIVAN: Building from what we've learned from our AdPath case studies, and particularly on retargeting, one of the things we think we can uniquely leverage is video advertising--even though everybody knows the amount of video on the Web is expanding rapidly, along with the numbers of people viewing video. Clearly big-brand advertisers are excited about it, including categories like fast food and CPG which haven't really moved so deeply into online before. But what's still less clear is the value proposition, given that the costs of a 15-second video ad online are nearly as expensive as TV. For video to really reach its potential online, it's got to be shown that it can target the right customers.

BI: How does video change the tasks of targeting?

SULLIVAN: There are some definite challenges in measuring video targeting. The brands that have been the biggest TV spenders and are now coming online, like fast food or package goods, aren't going online to sell hamburgers or soda or beer on the Web directly. But they do have promotions that can work in a direct response context.  So the way promotions are framed--and figuring out the way to test them--will be critical. That's where creativity has to ramp up.