The data-wonk founders of Media6Degrees (M6D) founded the company in 2008 on a simple premise: that by tapping the social graph of MySpace, they could parse custom digital audiences, or "tribes," for marketers trying to find buyers for their products. At the time, more than 100 million people had listed their likes and loves, dislikes and pet peeves on MySpace, and M6D sought to create algorithms to identify those most likely to "take a brand action," whether that meant product purchase, increased reach or site visits.
It made perfect sense; in fact, on the surface it seemed as neat an equation as you’d find in Euclidean geometry. But while their system performed admirably for youth-oriented products like cellphone plans and SIM cards, Puma and DoubleTree and Choice hotels, it didn’t work as well for more mainstream brands. Part of the problem was that MySpace, Hi5, MyYearbook and other social networks skewed to a younger demographic. But the bigger issue was that MySpace was deflating while Facebook was blowing up—but Facebook didn’t share data (today it does, but only to a limited extent).
This could have spelled the end of M6D, but instead it led the founders to their "a-ha moment." They realized that targeting data based on any social graph was too narrow. There were billions of other data points or brand impact motivators that ultimately prod a person to take action. And their algorithms had the power to observe the billions of motivators, translated into data points, that help shape a consumer's journey from attraction to purchase. Once they spiked their algorithms with these data points, they got far better results.
So in 2010 the company made the decision to move its core competency to the Web—where there are billions of people to observe. It was a pivot born of its time and an illustration of what can happen when math meets media. Five years earlier it probably wouldn’t have been possible. "You didn’t have the ability to work with data at the speed with which we can work with data now and the infrastructure to be able to serve ads in real time and deliver that decisioning and have it play out in real time," says M6D CEO Tom Phillips, a founding publisher of Spy magazine who graduated from Harvard with a BA in applied mathematics and has an MBA from Stanford.
Here’s how it works: M6D observes a client’s customers by deploying a cookie to track each person to, say, book a ticket through an airline’s website or configure a car at a dealership or auto manufacturer website, reporting back every page within M6D's partner network that customers have visited (M6D tracks some 100 million URLs). Say a sneaker maker signed on as a client and wanted to drive customers to its site to learn more about new products or get them to make a purchase. MD6's algorithm first models the behavior of proven customers of the sneaker maker, then it looks for people who are taking those same journeys across the web but have never bought these shoes. Those people will soon start seeing ads for these sneakers as they visit various websites.
This is where the company’s chief scientist, Claudia Perlich comes in. Perlich, a former IBM researcher born in East Germany who says she "loves" to work on real-world applications of predictive modeling and data mining, boils down the 100 million people in M6D's database to locate the people "on an accelerated path to conversion." To accomplish this, she deploys the company’s "prospect engine" to comb the Web. Unlike behavioral targeting firms, though, she isn't looking to group people based on their interests. Rather, she observes massive numbers of people on their online journeys because she has found that those who do similar things are often implicitly linked to a shared interest.
M6D then bids on available ad inventory using a real-time auction system, seeking to target the right customer at the right time—the ones in that top 1% of 100 million whose online journeys best match those of past customers. Depending on the strength of the brand signal, a person who Perlich identifies as someone whose web habits indicate they would be more likely to act on an ad is at least four times more likely to take action than a random set of people who see that same ad.
But don't just take her word for it: M6D now works with some 300 brands, including Allstate, Amex, British Airways, Disney, Hertz, Hyundai, and Orbitz, and claims that more than 80% increase their spend with the company quarter over quarter.
M6D, which has raised about $24 million in venture funding since its 2008 launch, is not the only ad marketing firm that wants to get to know you. A competitor, Acxiom, has amassed a half a billion dossiers to provide what it calls "a 360-degree" view of consumers. In contrast, M6D claims it doesn’t connect a person’s identity to the information it collects, nor does it capture or store any information about a browser beyond its anonymous visit patterns. Face it: To M6D, you're just a bunch of numbers—and they wouldn't have it any other way.
If you don’t want to be tracked, however, you can delete your cookies. (Here’s how.) But registration information you’ve provided to sites you visit will be deleted so you’ll have to re-register, plus pages will likely load slower.
It’s that usual trade-off between complete privacy and convenience. Which will you choose?
Adam L. Penenberg is a contributing writer to Fast Company and a journalism professor at New York University. Follow him on Twitter: @penenberg
[Image: Flickr user Leslie Kalohi]