When web ads are custom-targeted to the viewers most likely to be interested, powerful possibilities open up. Such targeting can be achieved with data mining, the science of uncovering meaningful patterns in masses of records (in this case, the traces of consumers’ online activities). For years, big Madison Avenue agencies couldn't achieve such viewer targeting, especially while keeping consumers’ information anonymous to preserve privacy. Media6Degrees filled this vacuum, thanks in large part to groundbreaking research in machine learning by a Stern Professor and computer scientist Foster Provost.
Coriolis Ventures, a longtime dot-com investor with deep relationships in the advertising world, incubated Media6Degrees and recruited Provost. He devised a breakthrough method by which computers automatically mine billions of web-page views for connections among Web surfers, while keeping all identities—and browsing behavior—private. Media6Degrees works from the premise that consumers care about the same things as people to whom they’re connected--the proverbial ‘six degrees of separation.’ Early in its incubation, Media6Degrees hit a wall modeling consumers into clusters because the scientific literature did not deal with mining this sort or scale of data: billions of records, tremendous variety and nuance in consumer connections, and the need for great speed. A paper Provost had co-authored in 2006 seemed to hold the key. Provost normally refused consulting offers, but this was different. “It was a great problem to solve,” he said. In 2008, he did.
Media6degrees raised $1.5 million from industry angels, then $9 million from U.S. Venture Partners and follow-on funding from Venrock. Late-2010 brought $18 million more in growth capital. From zero, it has mushroomed to 75 employees and $20 million in revenue (set to more than double in 2011). About 160 major brands use Media6Degrees technology. Provost still advises, part-time. His former doctoral student, Claudia Perlich, is chief scientist; Andrew Pancer, the COO, is a Stern MBA; Provost student Brian Dalessandro is head data miner and three other NYU grads work there as well. “They’re the smartest data wonks on the planet,” says Alan Murray of Coriolis. “We lean on the machine-learning and statistics guys at NYU for their great wisdom.” Provost returns his compliment: “Having great entrepreneurs is really the key.”