Beyond Re-Targeting: Targeting As Tailoring
Beyond the bells and whistles, a fundamental premise of Web 2.0 is that cultivating user engagement, interaction and community are the keys to building Web site loyalty. Yet, as Roy DeSouza, CEO of the Zedo Ad Serving network explains below, when it comes to serving ads, far too many Web 2.0 publishers fail to use what they know about their audience to deliver relevant, engaging advertising.
BI: For a majority of publishers today, behavioral targeting primarily means re-targeting visitors to certain kinds of Web page content. What lies beyond that?
DeSouza: As it’s conventionally practiced, behavioral targeting is in essence re-targeting, following a prospect from one content page to another to re-serve an ad. Experiments with that, by and large, have been very encouraging--but the amount of knowledge you actually have about the consumer you’re re-targeting is very sparse, really. There’s still a huge gap between the information publishers actually have about their visitors and the information actually used in behaviorally targeted ads. Publishers know more and more about who is onsite and what they’re interested in--yet that data so far hasn’t translated into improvements in relevancy and focus.
BI: Zedo’s new Profiling and Behavioral Targeting for Publishers seems especially geared to Web 2.0 sites. How does it address these sites’ needs and resources?
DeSouza: One of the really basic and critical things about so-called Web 2.0 is that Web 2.0 sites have a far wider and richer set of data at their disposal via their user profiles, which not only give you core demographic information but information of a psychographic nature, such as main interests and hobbies. Currently much or most of that information is just wasted.
Web 2.0 publishers are in fact leaving lots of money on the table by so poorly utilizing the data they have. So we’ve built a platform that allows publishers to construct far more detailed models of their customers based on profile information. We overlay data including age, gender, zip code, education and interests atop page content.
BI: Could you give an example of how that will work?
DeSouza: Well, there’s targeting and then there’s tailoring. Behavioral targeting as it’s mostly used now will locate two customers--customer A and customer B--and identify both as spending a lot of time on auto sites, say. So it will serve each of them auto ads as they move around. Which of course is a lot smarter than just serving them run-of-site ads. But it doesn’t give you much insight into whether you should target them with an SUV, a luxury sedan or sports car. But if you know customer A is a 24-year-old male snowboarder who makes $40 thousand a year and customer B is a 38-year-old female physician who makes $125 thousand a year, you have a lot more visibility into exactly which car ads you’ll want to serve.
BI: So increased ROI for advertisers leads to increased monetization for publishers?
DeSouza: From a publisher point of view, this is a way to tackle the biggest challenge of Web advertising going forward, which is the proliferation of new, under-utilized, ‘sub-premium’ inventory. As they become more Web-savvy, consumers clearly are spending more of their time away from the most popular home pages, but the obvious problem of untargeted long-tail pages is that there is never enough of a critical mass of users on any one page to justify high CPMs. So we have the ironic situation evolving where there’s increasing advertiser demand for quality inventory and oversupply of untargeted inventory. Publishers risk depressed CPMs if they don’t adapt, and adapting means better utilizing information.
BI: What goals do you have for your BT practice going forward into 2007?
DeSouza: So far behavioral targeting has focused on one side of the marketing equation, which is to more efficiently predict which consumers will be more likely prospects based on where they’ve clicked or searched or browsed before. We think there’s another dimension. Publishers can use the profile data they have to better forecast which pages of inventory they have that are particularly ripe for better utilization. Just as they can better segment their users, they can better segment their inventory as well.