Roundtable: Decoding Taste Through Social Graphs

NEW YORK – Whether users are shopping, browsing, or talking about their interests online, the success of the social graph depends on how well companies can interpret the results and use them to predict behavior, said experts at the APPNATION NYC Mobile + TV 3.0 Summit in SoHo this afternoon.

In a roundtable discussion moderated by TastemakerX CEO Marc Ruxin, some of the Web’s best curators discussed ways to collect, aggregate, and interpret consumer data. Panelists included Meebo president Martin Green, Percolate CEO Noah Brier, and Tumblr VP Andrew McLaughlin.

They discussed the merits of the major platforms: Facebook, Twitter, and Google+, and offered their thoughts on emerging networks like Tumblr and Pinterest.

“The Facebook social graph is composed of social connections between people who know each other,” said McLaughlin. Like Twitter, “Tumblr doesn’t require any kind of real-world identity. Our social graph tends to connect people based on interest.”

Interpreting interest is another story. On Tumblr, “Taste is expressed through likes, reblogs, and to a lesser extent, follows,” McLaughlin explained.

“Facebook is much harder to parse because you are less likely to like things that you like and more likely to like the people that you know and the things that they do,” McLaughlin added.

Noted Brier, “Facebook is trying to make that interest graph more relevant, but being able to predict things that people will be interested is where the majority [of advertising dollars will be spent].”

The panelists disagreed on how well Google+ will be able to compete in the social matrix. Said McLaughlin, “it seems to be less useful because it’s more chaotic and the population of users is much smaller.”

Green disagreed. “It’s not the size of your user base, it’s what you do with it,” he said. “People are going to Facebook for things other than buying products. With Google search, they can link interests between people who don’t know each other. They can – and do – use that information to tune search results.”

Surprisingly, the all-male panel had good things to say about Pinterest, which is said to have a user base of 60 percent women who account for 85 percent of the page views.

“I think it’s for real,” said Green. Rather than relying on a social graph like Facebook’s, Pinterest’s data seems to come “through human beings self-selecting what they’re interested in.”

The image-based design doesn’t hurt, either. “People browse on Pinterest like they browse on catalogues,” Green said.

For Pinterest to succeed, “You need to understand what’s in the photos,” added Brier, “and how they do that will be interesting.”

Another important aspect of analyzing the social graph is the “‘verticalization’ of interest,” said Ruxin. “Tumblr blogs do this better than Facebook and Twitter,” because bloggers tend to focus on a single topic rather than everything that’s going on in their lives.

The company only recently started selling ads, but “The pitch that we’re making about Tumblr for brands is that the ad unit is the blog post,” said McLaughlin. “In a couple months we’ll have some data.”

For now, he added, “We’re tring to figure out how well we can interpret the interest graph we’re sitting on.”