Abstraction

I’ve always thought that a real evolution of advertising is moving from an inferred model to a interaction model; that is, instead of inferring qualities about a person (based on demographics, buying habits, etc.), it’s much better to involve people directly.

I did a interview recently with Light Reading that touched on this topic.

http://www.lightreading.com/document.asp?doc_id=172485

Salient details to save yourself the jump:

“One of the most powerful aspects of subscriber profiles is maintaining a subscriber’s real-time context, and using that context in concert with personal preferences to drive policies across networks,” says Chris Hoover, Openet’s director of Product Management. “In this sense, the subscriber profile forms the foundation for next-generation interactive services. These are services customized based on an aggregate understanding of who the subscriber is, what the preferences are, and real-time context in which the subscriber is moving – without regard to network or device.”

Openet’s Hoover points out that, historically, advertising has tended to be targeted on the basis of inferences made from proxy data – basic demographics, for example – rather than positive identifications of definite interest. Such interest identification is near-impossible to do indirectly, as it is difficult to infer that a subscriber is, say, allergic to cats, or avoids alcohol [ed. note: or avoids alcoholic cats...], or is thinking of buying a new car. Behavioral targeting based on the inferential analysis of past behavior helps, but there is always a risk that the inference will be wrong – or even offend.

“But there is actually a new opportunity, which is to bring the subscriber into a bidirectional mode,” Hoover says. “By including the subscriber as part of the network profile – essentially, the subscriber as an active network element – the profile can be much more accurate. This isn’t, however, a profile that is built solely by the subscriber, but rather in concert with the subscriber, and including inferential data.”

“But there is actually a new opportunity, which is to bring the subscriber into a bidirectional mode,” Hoover says. “By including the subscriber as part of the network profile – essentially, the subscriber as an active network element – the profile can be much more accurate. This isn’t, however, a profile that is built solely by the subscriber, but rather in concert with the subscriber, and including inferential data.”