Bringing ISPs Into The Behavioral Mix

by Phil Leggiere, Wednesday, Nov 14, 2007 1:45 PM ET

Once the vital center of the Internet eco-system, Internet Service Providers are now more frequently looked at commodities, marginal if not entirely irrelevant to the emerging future of Web advertising. Yet, as Bob Dykes, CEO of NebuAds explains below, ISPs are a crucial, still largely untapped, component of a behaviorally based advertising infrastructure. Behavioral Insider: How does NebuAds work with ISPs in behavioral targeting? Bob Dykes: We started out with a very unique orientation. The challenge we set ourselves was to add value to ISPs and publishers of currently untargeted inventory, which includes about three-quarters of all current ad impressions. ISPs have been a neglected aspect of online's evolution over the past several years. But the fact is the depth of aggregated data they have to offer, anonymous data, is an untapped source of incredible power. In an era when cookie deletion is becoming more and more common, it's also more reliably continuous data. BI: How does the personalization and behavioral methodology you've evolved differ from what most of us are familiar with? Dykes: The conventional approach to behavioral targeting has been to place cookies on specific Web sites or pages. We've gone about it in a very different way. We place an appliance in the ISP itself. Therefore we're able to get a 360-degree, multidimensional view over a long period of time of all the pages users visit. So what we're really talking about for the first time is a truly user-focused, though still anonymous, targeting, taking the totality of anonymous behaviors rather than just a subset of sites on a network. BI: It seems at first glance this could open up a Pandora 's box of privacy problems. How do you avoid that? Dykes: We've had the advantage of coming onto the scene relatively recently when concern about privacy was growing. So we've architected our targeting platform from the ground up to addressing privacy concerns. For one thing, in addition to being committed to only using anonymous data, we have developed a privacy guidelines list which includes providing consumers with robust notice and choice such as instruments for opting-out of ads. BI: So tracking is not by individual identifiers? Dykes: We don't track individual consumers. We set up a taxonomy of commercial interest categories, for instance, ‘Porshe SUVs,' and simply keep track of content interests. We exclude data from sensitive areas, say sex sites or HIV Drugs. By anonymous we mean we collect no personally identifiable email addresses, last names, home addresses, social security or phone numbers, financial or health information. The kind of data we do aggregate includes Web search terms, page views, page and ad clicks, time spent on specific sites, zip code, browser info and connection speed. BI: How does the granularity arise? Dykes: Within this vast universe of information we create a map of interest categories, beginning with the widest definitions, auto, finance, education, what have you. But within those we can provide far greater granularity. So if you're talking about auto, we can drill down into particular interest segments, say SUVs, luxury cars, minivans, and then even to particular brands or models. Within the interest category of travel, we can identify consumers interested in learning about Martinique, the south of France or Las Vegas. Most interest categories age fast, of course, so one very critical component of analysis is identifying optimal cycles for reaching customers in each area. For flowers, say, there's a cycle that's at most hours in which to reach an interested consumer. With auto, obviously, the cycle is far more complex and includes several phases stretching out over days or weeks. By having a far more comprehensive view, from an ISP rather than site by site vantage point, we can identify from the highest level to the minutest level where consumers are in the funnel. BI: What's the key advantage for advertisers? Dykes: The advantage of having a more granular picture of user interests for an advertiser is that they have the capability to adjust and vary creative to suit customer interests. That's an incredible opportunity that often takes brands and agencies some time to really wrap their heads around. When we work with media buyers we educate them that targeting isn't only about serving your message to the most appropriate consumer prospects, but about serving those prospects the most interesting and relevant creative to get those messages across. It can be as simple as an iced-tea advertiser knowing whether and when to serve a creative using snowboarding or windsurfing. A more complex example would be, say you're an advertiser of high-end clothing and you're serving an ad to a woman who's very interested in high-end clothing and accessories and is on MySpace. Instead of just serving the same ad for Fendi handbags several times, you can serve a series of ads with different varieties of handbags. The notion of retargeting as conventionally deployed is very limited. You need to mix up different types of creative or it gets boring. Another advantage of having a more comprehensive set of data points is that advertisers can learn to better mix their ads in an optimal sequence to match consumer interests according to where they are in the buying cycle. They can learn to know when to serve a first impression which is very soft-sell, when to provide more detail and invite more engagement, and, finally, when to really try to drive a transaction. This not only provides more variety but integrates creative, marketing strategy, consumer interest and the product sales cycle in much stronger alignment.