• by Steve Smith , Staff Writer @popeyesm, July 17, 2009

Cracking the social media monetization code has become one of the main obsessions of behavioral targeting in the past year. The data produced by social networks is just too juicy, too intimate, too valuable for any self-respecting data miner to resist.

In these columns we have seen how MySpace and Facebook both have tried to leverage all of that valuable user activity in their networks for more effective targeting. Other ad-targeting technologies like Lotame and Media6 have been using the social graph people develop inside of social networks to segment and target them when they are elsewhere on the Web.

And we have only begun. The great promise of social media for advertisers is that it represents such a rich font of relatively uninhibited exchanges among people, troves of intentionality, a mosh pit of sentiment and real records of what people do. We are only scratching the surface of ways to leverage this new kind of virtual conversations.

And so at the July 30 OMMA Behavioral show in San Francisco we dedicate a panel to the intersection of social media and behavioral targeting, "Making Social Media Behave," which will be moderated by one of the old pros of Web analytics and behavioral targeting, Anil Batra.

We invited to this panel Lisa Rutherford, CEO of the novel start-up TwoFish, because her company focuses on a specific kind of emerging online community behavior that is as new as it is promising -- virtual economies. In virtual worlds and online multiplayer games, members are upgrading their avatars, gifting one another, accessing new game levels, or just buying weaponry with special powers through virtual currency systems.

Generally the economy requires some up front cash payment that translate into digital dollars. In-world purchases of virtual goods and services can take place at any scale and even get redeemed for real-world goods over time. TwoFish not only services virtual economies but also collects what Rutherford calls "perfect data" about transactions in these worlds. Who bought what +4 Magic Sword in what world for how much? What happens to that sword over time as it gets traded across different players? What leads players to buy the sword and at what time of day? When masses of such data are compiled and sliced in just the right ways, they may render both worlds makers and their advertisers unique insight into the consumer and how to market to them.

In many ways the virtual worlds let a company watch people shop over time and across many different products at a level of detail that mimics a real world people meter. It can see, for instance, just what the player or buyer sees right before making a purchase. "We fundamentally believe that what is exciting about the virtual part of the virtual economy is that we can collect very perfect data," says Rutherford. "We can help you convert your users from non-paying customers to paying customers. We use post-purchase data to help optimize the game/world and revenues." Virtual worlds makers can change their pricing structure, the scarcity of items, the timing of offers in world to accommodate the buying cycles and even the item usage patterns of members.

Of course, the same knowledge can help advertisers insert themselves into the virtual worlds and their economies. We are only beginning to see how this sort of online behavior can be leveraged and used both in world and outside of the virtual space. "Think of it as a transactional graph as opposed to a social graph," says Rutherford. "Take a real-world analog. I am a female in her thirties, married and living in Palo Alto. You know I go to the gym, to local restaurants. I shop. I go to work. That is all fine. And that is equivalent to online behavioral targeting. But wouldn't you really like to know that I only buy organic fruits, that when I go to the gym I take yoga and pay this much for the class? And when I go to a restaurant I often order chicken. That gives you interesting information in the real world that might make you send me a health magazine as opposed to Us weekly. We can do all of that in a game."

Tracking virtual purchases within a virtual world can help a content provider know when and from where the user prefers to play or shop. If a racing game player likes to come in-world from the office PC at noon for a few laps of play but comes back on at 10 p.m. to upgrade his car and swap in new tires, the patterns can reveal a great deal about the user. The goods and the logon experience can be contoured to make the world more seamless and responsive. A quick logon screen might meet him during the day so he can play quickly, but an advertiser will know the ads are wasted on this user in this day part. In the evening we know he is in shopping mode and that is where to concentrate the offers.

One of the first tests of sophisticated content and ad targeting based on virtual world behaviors will come in a few months when the "Star Fever" massively multiplayer online game launches aimed at celebrity-focused women. Players have starlets that they manage and try to propel to fame. Like a Paris Hilton Tamagotchi, the starlet requires a steady supply of lattes and clothing upgrades. "Star Fever" will be integrating advertising into the game and Rutherford says that company plans to push data mining and behavioral targeting farther than any MMO yet. "It is going to be an eye-opener in terms of how to use data to make a better experience and ways of working with advertisers."

How virtual economy behaviors translate into real world habits is hard to say for now. But the publishers of these virtual worlds ultimately own all of the data and the insights a system like TwoFish's generates. Even in social networks like Facebook, where gifting and virtual goods are starting to emerge, these economies embody a set of behaviors, preferences and intents that are something less than pure fantasy -- if also something other than real. Once publishers have these fascinating profiles of users' in-world behaviors, then conceivably they could use them to tag and follow that user elsewhere to see how responsive they are to marketing pitches elsewhere.

Does the choice one makes to buy the Seattle's Best Latte in a game world (rather than Starbucks) parallel a preference in the real world? We don't know yet. But it looks as if we are about to get the data that will tell us the answer.