Over the past year the online video space has seen an explosion of new video content, new consumers of that content, and new advertising inventory. Ergo, theoretically at least, expanded content, plus users, plus inventory, SHOULD, by all measure, equal rich expansion of opportunities for behavioral targeting of in-stream video. In reality, that's not really been the case, at least not yet. Turning the video usage boom into a scalable targeting opportunity means the industry must seriously address one big elephant in the room, the lack of third-party ad serving for in-stream ads, Tod Sacerdoti, CEO of BrightRoll, explains below.
BI: What is the current status of video targeting, particularly behavioral targeting?
Tod Sacerdoti: Enhancing targeting has been the focus of most of us in the industry, but there's been frustration because we're still far from where we want to be.
BI: Why is that?
Sacerdoti: The first thing you need to distinguish is what networks are talking about when they're talking about video behavioral targeting. There's a big distinction between serving a video ad within display versus in-stream. Often, indeed, blurring that distinction is a way of avoiding the issue altogether.
If what you're talking about is video within display, then behavioral targeting proceeds on the same basis as with conventional display inventory. But if you're talking about in-stream video, what you'll find is that at least 80% of video ad inventory is targeted only on a URL level. The fundamental reason is that the golden 15, so-called -- the broadcasters, cable networks, portals, YouTube and Hulu -- do not enable third-party ad serving. That's the elephant in the room. Until it is addressed, the prospects for in-stream behavioral targeting are minimal. Until we get to that point, which I think is inevitable. I do know that if every impression were actually served today, we could do user-targeted video.
BI: Given those ongoing bottlenecks, where is targeting in video currently working or gaining significant traction?
Sacerdoti: The most interesting type of targeting we've been working with has been combining in-banner and in-stream. In one campaign with a large insurance client, they saw success retargeting users exposed to an in-stream ad with an in-banner one.
We are seeing significant indicators that retargeting using video assets is following a comparable path to what we saw in the display space three of four years ago. An example is a large IT firm which has enormous site traffic, well over one million unique users a month. The client said that they wanted to target any "IT professional" users we could locate in an in-stream environment. This approach is now common in display, but it's still a new thing in-stream. I think a year from now it will be much more common.
BI: There's a lot of concern right now about prices for video inventory falling. What's your take on that?
Sacerdoti: Online video CPMs have been falling in aggregate at about $1 per quarter. We think that's a very good thing for the category. Because online video, however valuable we in the industry all feel, and know it is, is simply much more expensive than TV. It's hard to argue for online value as being 50% more valuable than TV. But I'm very comfortable arguing that online video is equal in value to TV. I think that when online video CPMS have fallen about another 30% to 40% in pricing, we'll finally be at the one-to-one level where we can make that case.
BI: How might better user targeting affect that dynamic?
Sacerdoti: The basic dynamic is that if you look at the evolution of targeting on the Web, the reason pricing drops is there's a continuous process of cutting out waste. With almost infinite amounts of inventory available, the value of unwanted, untargeted impressions continues to fall. But at the same time, advertisers are willing to pay more for the impressions that, through precise targeting, have value.
So what I predict we'll see evolve is a three-tiered model of inventory. A sliver at the top of the most premium quality video content, ESPN and CNN for instance, will command top price. A huge bottom tier of undifferentiated inventory will trend toward zero value, and a middle tier of inventory which is enhanced by targeting data will rise in value. Think of inventory such as Yahoo music videos: by itself it probably will have limited value. But if you could locate auto-intenders downloading music videos, that greatly enhances the value.