• by Phil Leggiere , May 21, 2008

Social media is by all accounts more about making connections between people than about publishing content in the traditional paradigm. Yet, while this is no secret to either publishers or marketers, nearly all social network targeting strategies have remained strictly focused on leveraging profile data tied to individual behaviors. What they've continued to miss, is the social connectivity piece, explains Jim Calhoun, CEO of PopularMedia in the first of a two-part conversation.

 

Behavioral Insider: What drew Popular Media to focus on social networks? What are the unique opportunities of social networking for target marketing as you see it?

Jim Calhoun: The company you keep says a lot about you. If we know a little bit about who you're connected to - even if we don't know much about you, or your past behavior - we can paint a surprisingly accurate picture of you, your tastes, response profile, and more. You consult your friends to inform the decisions you make every day. Like it or not, they influence who you are, what you eat, what you wear, how you act, what you watch, what you read, where you go online and much more.

Our company is at the forefront of an emerging type of targeting that's based on how people are connected to one another. We develop technology that analyzes connectivity, along with other factors, to make predictions on what type of content we should serve up to any given individual in a social media program. We call it 'graph-based targeting.' Graph in this sense is just shorthand for the social graph, a word used to describe the social bonds that exist between all consumers.

As the technology progresses, we'll be using socially connectivity data to enable publishing partners to predict and inform exactly what content, offers, experiences or ads they should present to any given individual. Your online experiences five years from now will be greatly informed by the company you keep; the tastes that surround you; the actions of your friends and neighbors.

BI: Could you elaborate a little more on what the graph approach is?

Calhoun: The graph-based targeting approach differs from traditional online targeting, which is all about collecting more and more data about individual consumers. Ironically, even the big social networks have fallen prey to the allure of data-driven targeting. They essentially sell advertisers the ability to place ads based on highly specific profile data. It's interesting, but there's nothing really predictive about that approach.

As a result, ad performance on social networking sites suffers because marketers don't have the time to craft thousands of variations of ads and landing pages required to take advantage of this rich type of hyper-targeting data. It's not scalable because it's impractical from the buyer's point of view. The production, strategy, planning and execution hurdles would kill any lift you might be able to achieve. Graph-based targeting takes the best of profile-based targeting and brings scale by taking existing ad formats and presenting them on a predictive basis to people we can predict will find those ads relevant with a shocking degree of accuracy.

BI: What are the biggest challenges or obstacles to effectively growing a targeting platform in this medium?

Calhoun: Understanding how consumers are connected is a big barrier to entry in this field. A few players have great social connectivity data, like Microsoft, AOL, and Yahoo - I'd argue they have a bigger opportunity here to create value than even leaders like Facebook or Myspace.

On one hand, the major players are all trying to grab headlines and 'out-social' one another with feature announcements, standards grabs, etc. while they figure out how to get their behavioral targeting technologies to sift through 500 billion cookies to deliver a .005% lift in a banner buy on remnant inventory. When it comes to targeting, the major players are pretty much up to their eyeballs sifting through unworkable amounts of clickstream data, or they're paralyzed trying to solve lucrative but completely tactical challenges, like how to gather gender data and append it to a cookie. All of these potential players are distracted with the feature side of the social media phenomenon. If the major players put pen to paper and actually did the math, they'd quickly realize that they should be investing heavily in the rich predictive targeting opportunities social connectivity data offers. But predictive graph-based targeting algorithms are geeky stuff that lives below the crust, where it's unlikely to grab headlines in naval-gazing tech pubs.

BI: How is or can predictive analytics based on behavioral data be deployed to improve targeting of social media campaigns areas?

Calhoun: Behavioral targeting can play a big role. It's great at giving insight into a person's interests and intentions -- at least for a snapshot in time. As that data gets added to the social graph, you start learning more about not just the people you observe -- but you now can predict things about people you've never seen before, since you know a little about the company you keep.

Until we start marrying behavioral with graph-based targeting, there is no real advantage to using behavioral targeting in social media campaigns or on social networking sites. You'll get the same amount of lift as you would (using BT) with any other sort of untargeted inventory.

The bet Myspace and Facebook made early was that profile-based targeting (detailed segmentation) would offer better results for marketers. It doesn't, and it's too cumbersome for marketers to take advantage of at scale. BT is a better solution for Myspace and Facebook, until they get their act together when it comes to graph-based predictive targeting for content and ads. When you combine predictive algorithms that integrate all these factors -- profile data, behavioral data, and connectivity data -- you can achieve stunning lifts in relevance and accuracy without forcing media buyers to do something wacky or out of process.