Marketers are buying audiences “in every imaginable way right now, “ according to Roza. While Datalogix has its roots in direct mail analytics, its online work began five years ago in the open Web, with ad networks and then trading desks. Over the last two years, through a partnership with Facebook, the company has been able to add rich offline purchasing data in the context of the social network.
Roza is seeing mobile, video and social becoming viable venues for premium audience targeting as well as measurement. “You’ve got to extend your audience, ensuring you’ve got enough volume and the right people. With the right approach, marketers can make things more economical for all stakeholders. If you can reach the same caliber of audience on the sites as you can off the sites -- and prove that -- it ends up a win all the way around,” says Roza. “Plus, substantiate that with sales metrics that say ‘Yes, I moved more cars for my luxury brand,’ as well as, ‘I’ve got the same 6-figure demographic, plus I was able to deliver this at a 20% cost reduction.’"
I was interested in learning from Roza his feelings about the intent versus audience question. There’s a perception, particularly in regard to premium display, that intent does not factor in heavily enough, and that the cookies we use to target audiences simply aren’t fresh enough for ads to be relevant.
Roza believes in taking this on a case-by-case basis: Sometimes cookies trump intent, sometimes intent trumps cookies, and often the two are complementary. Intent, for example, is not a huge component for a consumer considering a premium ice cream purchase. In the travel vertical, however, it’s positively critical.
Elsewhere, the interplay of intent and audience is more complex. “Let’s take automotive as an example,” he suggests. “Premium auto endemic sites have some of the highest CPMs in digital. However, with the right audience data, a slightly stale cookie can still be tremendously valuable. For example, if someone’s got a three-year-old Lexus and they typically buy a car every four years, every luxury automotive company on the planet should be paying 5x to reach that consumer right now, regardless of context. On the other hand, a broke 19-year-old college student checking out Lexuses on kbb.com, who drives an eight-year-old Honda, isn’t worth a penny to those advertisers.”
“I always get triggered when people say categorically ‘clicks don’t matter’ or ‘offline purchase data is irrelevant in an online world,’” Roza tells me. “It’s really dependent on the category or product, as well as the message you’re putting out there.” He illustrates his point with a hypothetical scenario about a new, improved P&G Swiffer. If P&G were launching a campaign in support of the improved product, intent would be unlikely to factor in. It would be far more critical to have information regarding who had bought Swiffers in the past, who had bought a competitor product, and the demographics of both parties. That, according to Roza, “is going to trump the hell out of any intent data.”
Ultimately, it really does depend on the situation. We make so many mistakes in advertising if we let our intuition do more than generate the hypothesis. With so much data and technology around us, there’s every reason to test our theories. “Test your theories in small, low-cost experiments, then iterate it,” suggests Roza. “I think that’s always the best approach versus having this religion around things.”