The usually unexamined premise of behavioral-based marketing is that personalizing content means ever more deeply scrutinizing what makes individual consumers unique -- the better to forge that proverbial one-to-one relationship between brands and consumers. As important as the drive to personalize and individualize content is, however, it sometimes neglects the fact that consumers as people are socially bonding animals. Which is to say, as Deborah Richman, senior vice president of Collarity, explains below: Paradoxical as it may seem, sometimes what makes individuals unique is their relationships to other individuals.

Behavioral Insider: Collarity has made a big effort to apply behaviorally derived data to the challenge of targeting video. What do you see as the biggest obstacles to targeting this channel based on conventional approaches?

Deborah Richman: We recognized several obstacles in conventionally targeted videos.

First, the conventional form of categorization for video content is based on predetermined static descriptions and metadata about the video, the actor and the content category.  Without relevant dynamic behavioral components, the "content" you're attempting to target is often useless background information.

Another relevancy obstacle becomes the keywords or comments attached by video producers/creators or viewers.  Unfortunately, when relying on biased producers or vocal users, there's very little organic segmentation of video by usage and overall user base.

Finally, individual users have very ephemeral and frequently shifting interests, which cannot be statically defined or pigeonholed based on a singular category of content they happened to have viewed.  Rather, the culmination of their interests needs to be taken into consideration for effective overall targeting.

Where we differ in approaching the issue of behavioral data is within our methodology. Collarity is not dependent on meta-tags, nor is it limited to what an individual does in isolation from other people on the site.

BI: What kinds of behaviors are leveraged, and how do you identify these more collective, micro-community behaviors?

Richman: At a publisher level, we aggregate 100% of click-stream data and generate community segments based on how a specific video consumer's interest connects with all other users at a given time. We track how video-viewing habits and passions constantly shift as individuals join and un-join what we like to call ‘implicit communities.'  With Collarity, video content is served based on each individual's video habits at a given time in relationship to ‘implicit communities' of interest they are associated with.

BI: What kinds of data in particular are you talking about?

Richman: With Collarity, at any given time segmentation is generated based on several criteria:  keywords; videos viewed; ads viewed; and the ‘implicit communities' of interest outlined above. By ‘implicit.' I mean communities that form dynamically and behind the scenes. For instance, traditional behavioral targeting might take note of the fact that a particular user was looking at video that had a "car" description on it.  That user would likely be tagged as a "car enthusiast" and served car ads. This individual may have no interest in cars beyond the mere video that they were watching at that moment.

Collarity looks at users realizing that their interests are dynamic and changing. For example, let's say someone is watching ‘Smoky and the Bandit.'  The implicit communities that have formed may reveal a "Burt Reynolds" community, or it might be ‘'70s retro.'  It could even be ‘pick-up trucks'! What happens is that users see related video choices, based on available publisher archives that correspond with their own communities of interest. Ads are then targeted and served based on these interest groups instead of mere keywords.

BI: What kind of educational curve do you see going on at this stage of adoption of this approach?

Richman: The consumption metric is what is most important for media buyers to understand, since viewers targeted with dynamic implicit behavioral tracking show dramatically higher ad consumption.  Users look at more video content of all kinds, including ads that are viewed as just another kind of content, and in our experience they view twice as many banners.