JONATHAN STEUER: People love to pick on Nielsen, but they did the job people paid them to do, which was make numbers as consistent and comparable over a long period of time as possible.
The networks weren’t clamoring for new kinds of measurement. It wasn’t until OTT kicked in when the traditional TV industry worried about the value of programming across multiple distribution chains. That’s a set of questions everyone wanted to answer. Nielsen answers that from the content side, but they haven’t figured out how to answer it on the advertising side.
What do you mean when you say Nielsen hasn’t stepped up on the advertising side?
Content used to be a proxy for the ads. They were embedded in the TV program and most people watched live, so you were only on the hook to answer what audience was on a show when it aired.
Now, with programmatic TV, advanced TV and addressable TV, we’re finally getting digital-style ad targeting capabilities into traditional TV. If you’re used to a ratings world, it’s a challenge to move to impression-based measurement.
You’ve noted the limited footprint and very expensive inventory of addressable TV. What’s real today and what’s aspirational?
You can argue the numbers, but there are probably only about 25 million addressable linear households where you can put the ad directly into an ad pod. And another 10 million – 25 million, depending on how you count – where you can do addressable on demand.
And the way the two satellite companies work is different from how everyone else has implemented.
It evolved for an interesting reason. Unlike cable, satellite has no way to air local commercials because they send the same signal to the whole country. Cable operators cover specific geographies and each insertion point can send a different ad based on geography.
The satellite guys, as they deployed DVRs, built tools to use the DVRs to send commercials and store them locally on the devices. They developed a signaling mechanism to play out the one that’s appropriate for a specific audience.
So in building something that let them do local insertion, they ended up with something that let them do household-level addressability.
So there’s basically a bunch of different addressable technologies to navigate?
Cablevision is the only traditional cable operator that, because they operated in a single market in New York, has a homogenous back end.
But all the other big cable operators, and especially Comcast and Charter, grew by acquiring a bunch of small operators in a bunch of different geographies. That gave them a heterogeneous technology footprint. So doing household addressability the same way for everybody is difficult.
With Cablevision, the whole footprint is addressable. Comcast and Time Warner, which is part of Charter, have some addressability in their video and demand stack. They can insert ads on the server side.
Are the cable companies investing in addressability at all?
The vast majority of their revenue is subscriber revenue, not advertising. So when the technical obstacle toward doing household-level addressability was high, they didn’t spend money on it. They spent money on things that kept the subscribers paying. But with cord-cutting, now they’re making the investment.
To what extent does the AT&T acquisition of Time Warner change the addressable TV landscape? Do you get the sense this is something AT&T actually intends to be a part of?
It's certainly significant – more so for the long-term than any immediate impact – specifically as it relates to long-term potential of enabling addressable advertising to move beyond the living room set-top box to mobile devices via content-oriented application
How do all these different technologies affect Omnicom’s ability to execute an addressable campaign?
It’s annoying, but not impossible. The good news is that there are only four or five guys who do addressable at all.
But it’s still less than 20% of the total US. The bulk of our work is with national advertisers on big campaigns. As more addressable inventory comes online, we’ll need to manage this in a more automated fashion.
And none of that is what we call programmatic TV.
Once the automation component kicks in, isn’t that programmatic TV?
At scale, yes. What’s today branded as programmatic TV is remnant inventory from cable operators and local broadcasters. That’s getting rolled up by companies with technology platforms that can run viewership against targeting data. They try to package up the small amounts of inventory they have access to in ways that make it more data-driven. Then they put it in a platform that allows for some amount of automation.
What would a mature programmatic TV environment look like? A pooling of inventory across all broadcasters? Or a bunch of walled gardens?
We don’t know yet. There are three broadcasters on the leading edge cobbling this together on their own. They’re focusing on making their own advertising act like early ad networks. They’re packaging up their inventory in ways that are attractive to specific audiences.
From an agency perspective, we want to buy the way we do now: with direct relationships and a combination of our planning, targeting and data.
Does an addressable TV buy ever come into play during the upfronts?
Right now, there’s such a size mismatch that it doesn’t come up. The addressable buys come from paid TV, not from networks. Networks can’t deliver it. Addressable can’t be sold this way because it’s not even sold by the same entities.
So under what circumstances would OMG be doing an addressable buy? Are these experiments or do you have specific goals?
Addressable TV allows much finer audience segmentation. You pay the premium to reach your target, and hopefully you save money because you’re not hitting people with an irrelevant message.
Also, if you’re a car company, you want your ad to have people call a number to take advantage of an offer. It’s nice to segment that at the local level. But then again, you can do that with zone cable, too.
Is device- and person-level targeting real in the TV world?
TV is difficult to measure down to the device or person level. If you think of the mechanics of doing that, it’s really hard. TV is a social experience.
Are there mechanisms that aren’t too invasive that can detect who’s viewing?
Well, that depends on what you mean by “too invasive.” There are two classes of tools I’ve seen; neither are widespread commercially.
One uses audio content recognition that runs in the background on mobile phones. Axwave is one company that does that. Alphonso TV is another. They pick up on ambient audio and try to match it to audio signatures on TV shows.
The other – and this is where you wonder how intrusive is too intrusive – are a couple companies like TVision, which have camera-based measurement solutions. They do video fingerprinting via a camera, like the one on the Xbox Kinect or a webcam mounted to the TV set. That uses facial recognition to determine who was in the room and determines if they are happy or sad.
In defense of the potential creepiness there, it’s all opt-in, so no one is deliberately tunneling into your Xbox and stealing that data.
People have talked about Bluetooth and RFID sensor, but neither is ready for prime time.
If the FCC unlocks the set-top box, will that change the way you buy?
The idea of unlocking the boxes all of a sudden is a misnomer. There’s been a marketplace standard called cable card that lets consumers bring their own set-top boxes. It’s just a very cumbersome technology to deploy and use. Now they’re looking to deploy a software-based solution to make it less cumbersome.
The details of how they do that will make a big difference in terms of who controls the data that comes from those different devices. How it works will make a big difference. If it’s about operator MVPD-controlled apps on smart TVs, the MVPDs might remain in control of that viewership data. On the other hand, if it opens up to another ecosystem where the channel players are involved, the answer might be different.