![Friendship data: 33Across crunches the numbers [audio interview]](https://www.smartdatacollective.com/wp-content/uploads/2010/03/friendship-data-33across-crunches-numbers-audio-interview-860x484.jpg)
If you leave a comment on this blog post (and I hope you do), how likely is it that you’ll be interested in the most recent book I bought on Amazon? (FYI, Chess Metaphors–Artificial Intelligence and the Human Mind arrived just yesterday.) I’d say there’s an excellent chance that you would not buy that book. But even if 90% of you ignored an ad for it, advertisers would be thrilled if the other 10% clicked. In their world, a 10% click rate is akin to winning the lottery.
That’s the thinking that drives 33Across, a number-crunching start-up in New York. The company, in offices perched high above the rail yards west of Penn Station, tracks social network behavior of some 115 million anonymous Web surfers in the United States. If one of them checks out, say, an office chair online, 33Across can lead an advertiser to the people in touch with that person. Maybe they forward articles to each other or comment on each other’s blog. (Another company chasing the same market with a different approach is Auren Hoffman’s Rapleaf. According to a Mashable story, the company has data on 389 million people worldwide.)
Eric Wheeler
Yesterday, I stopped by 33Across and chatted with Eric Wheeler, the
co-founder and CEO.
Listen to the interview:
He says that in some categories, people’s social
contacts are more than five times as likely to click on the same ads,
and buy the same products, as the general public. This type of
friendship analysis is hot now, because with Facebook, Twitter and
other social tools, online humanity is producing torrents of friendship
data. Mining these relationships (while steering clear of a privacy
backlash) could finally provide those companies and others with a
social network business model. (I wrote a BusinessWeek cover story about this last year. And the possibilities for this data extend far beyond marketing. Controversial research even attempts to correlate obesity to patterns of friendship.)
These are early days for this type of analysis. 33Across can supply companies with people’s strong ties–people they contact on a regular basis. Most of us have about five to 10 of these. They can provide greater numbers of weak ties. If customers give them sample data, they can target people by age and gender. With time, outfits like 33Across will be able to sharpen the focus. What are the correlations among people who communicate more on weekends, or late at night? (Just guessing, but I’m betting that a liquor company would be more interested in people we talk to at midnight than those we deal with at noon.)
In any case, this type of analysis is useful for advertisers who want to scale their data. If 10,000 shoppers check out a product, the chances that they might be enticed to buy it are high. But how to do expand that list from 10,000 to 100,000? Easy, by contacting their friends.
I recorded a 11-minute chat with Wheeler. I’ll link to it when it goes up on Smart Data Collective. (Below, the rail yards below 33Across. That’s the Hudson in the background)
