• by Phil Leggiere , October 15, 2008

Behavioral targeting in the banner display ad arena has, so far in its still-young life, been driven primarily by a direct response imperative to use search and browsing data to locate where potential shoppers are in the sales funnel. Highly appropriate as this may be in direct response, the challenge of more brand-oriented video is to extend the emotional range of behavioral targeting to learn to target passions, Steve Mitgang, CEO of Veoh, explains below.

Behavioral Insider: We've been hearing about behavioral targeting in video for quite awhile, but so far it's been off to a slow start. What are the unique challenges to bringing behavioral into video?

Steve Mitgang: There are unquestionably some real distinctions between behavioral targeting as it's been practiced for standard online display ads and how it can be applied in a video context. On the whole, up to now behavioral targeting has been mostly about consumption of information of the kind you get from searching or browsing. The goal is to locate consumers at a deep place in the product funnel.

BI: So it's less of a direct-response motivation?

Mitgang: Right. With video you look at the passions underlying the consumer's activity. We called it behavioral targeting out of a connection to the conventional lingo, but really it's a question of passion targeting for brands -- and we feel video is more about brands than traditional direct response. The point is less where someone is in a shopping funnel and more about connecting with particular kinds of people. So the goal is to track how high-passion interests are expressed in engagement with specific kinds of video content.

BI: So content genres are your building blocks?

Mitgang: The more meta-data you could aggregate, the better. Genres are the building blocks, but there are many different ways to describe and, more importantly, understand how video contents are being consumed. For instance, someone is watching a specific TV show and knows it's a detective show  -- but what might really be the most relevant thing in the context of a particular user might be that a particular actor whom they follow  is in that show.

We sometimes describe our user base as the ‘YouTube graduates.' There's more longer-format content, and people who upload videos tend to be more thorough and detailed in organizing and tagging their video content. Another thing that's critical is that a systematic effort is made to filter noise out.

One big way we do that is by using the community to supervise the tagging process, by saying things like ‘this video isn't really tagged adequately or accurately.'

BI: So how are segments derived from there?

Mitgang: In the simplest sense, if you have video organized and tagged intelligently, you can follow content usage and interest very closely. If a user pulled up a video from BMW films and also watched Fast Lane Daily videos, those two things would qualify them as ‘Auto Enthusiasts.' We've got one of the largest software firms in the world as a client. They were interested, instead of just running their standard advertisement, in doing a very technical short film to run in front of people who were serious techno-geeks. They knew it would be a relatively small audience, but one that had intensity of passion for technology.

That would be an example of a narrowcast approach. But it could be quite a bit more broad-based. If you were an auto manufacturer looking to reach a younger demographic of auto enthusiasts who are a prime market for a sub $25-thousand car, you can reach them with online video more precisely and on a greater scale than with TV.

BI: How about the problem of brand risky placements, the adjacency issue?

Mitgang: It's becoming increasingly clear that if a brand wants to reach audience segments between 15 and 34 in a concentrated manner, the opportunity to do that on the networks is fast dwindling. The issue of risk about having your brand message with the wrong content still exists. But those risks are melting, as targeting and placement efficiency improves. There's more of a realization now than a year ago that wide syndication of video content is a natural model online, and you just can't get scale advertising only on endemic content like CBS.com.

BI:  What's your thinking going forward about optimizing ad creative and messaging in a video context?

Mitgang: Looking ahead, one of the focal points of innovation will be using video behavior to optimize ad placement, frequency and even creative format. Even more fundamentally, I think what's on the horizon going forward is a radical shift in how we think about ad units. Ad units won't have a "container" - meaning, they won't be your standard 15- or 30-second spots. 

I think to get a sense of what the future holds, you need to ignore that when you come to a video break what you may have instead of a 30-second ad is a series of options for sponsored interactive expression. So say you've been watching a Madonna concert video, and it's sponsored by Pepsi. When you get to a break, you might have the choice of buying a DVD or tickets to a concert -- or you might even have a widget application where you can dress Madonna.