Building on the Social Network's Behavioral Targeting

Emily Riley | December 07, 2007, 04:05 PM

It’s good to hear that MySpace is actually selling a serious chunk of advertising direct now, rather than through networks. According to a report on MediaWeek, 60 percent of MySpace’s inventory is now sold directly, 30 percent comes from Google, and only 10 percent goes through AdNetworks. The entertainment channels on MySpace are advertiser friendly, but the vast number of impressions on user’s profile pages is no more interesting than previously. For them to be at the 60 percent in-house mark, I am willing to guess that some of the direct response banner advertising that used to run through ad networks is simply direct now.
In order to really lock in big revenue growth in the future, MySpace may to need a breakthrough advertising system that makes the most of the long tail of low value impressions. Granted, MySpace was the first big social network to announce a new method of display targeting not so long ago, namely, categorizing influential groups into 10 or so large verticals. However, this is really the tip of the iceberg. Just because someone has a picture of a car on their page or says they like to travel doesn’t make them the ideal ad target. More recently, Facebook has grabbed the spotlight with their Social Ads, which allow for some level of ad targeting based on user generated content as well. Still, the targets are based on the networks someone belongs to and the preferences they state when registering. What I’d like to see is true UGC targeting that goes beyond contextual and looks more like next-generation behavioral. If the social networks make this play, they’ll go a long way to increasing their inventory’s value, but advertisers will still be limited to the one website. If behavioral targeting does evolve to include UGC across vast numbers of sites, it might need to be a behavioral or other banner network that does it. Many have some capability to monitor UGC and target against it, but no one has decided how to make it scalable or how to categorize such wide variations on themes.