Commentary
Addressable TV: If It's So Smart, Why Isn't It Rich?
- by Ted McConnell , Featured Contributor, December 1, 2016
The big reason for that disconnect is that addressable boxes show mostly linear TV. Ads inserted at the network level don’t benefit from addressability, and don’t count as addressable ad spend, even though they are showing on an addressable box.
The squirrely underbelly
What makes planning harder is that in most, but not all addressable, you can’t use content as a surrogate for audience. You can’t buy presence on a program.
This is not only a technology issue. It’s a channel conflict, resulting from the simple fact that carried networks don’t want their distributors selling against them. Instead of selling an ad in a program, the distributors drop it on top of a broader denomination such as a daypart.
advertisement
advertisement
For example, ad insertion for satellite networks, which comprise many of those 50 million households, is simply an overlay, at a wall-clock time. If the TV was on, the ad can display. In these networks, each set top controls its own ad serving by dynamic rules transmitted as data over the satellite network. All available creatives are stored locally in the set-top box.
Free of the constraint of content-based insertion, it’s easy to show an ad to any household because it does not matter what they are watching.
Addressable and programmatic
Denizens of the programmatic domain swim in the concepts underlying addressable TV all day. For example, a device identity (like a cookie) is anonymous, but points to deterministic attributes (derived, in this case, from the relationship between the network and the household). Privacy is complete, but audience attributes are known per device. Despite this, digital folk are often unaware of what is technically possible. Custom audiences? On-ramping? All possible.
What is addressable? Broadly, it’s a communication system in which end points can selectively talk to each other. Almost anything with a black box is technically addressable (cable, satellite, streaming, etc.) — but addressability, by itself, does not assure any media capabilities. To know what’s possible, you have to work with each supplier. That, by itself, chafes against easy scale.
At its core, addressable TV fits with programmatic delivery due to the relative certainty of being able to put an ad in front of a defined audience, and only that audience. Dish offers a private marketplace for addressable TV that presents ad opportunities, impression-by-impression, available to demand-side platforms. This is almost like digital video, except delivery data is household-level, and there are no clicks. For brands, these departures should be trivial compared to the benefit of ultra-targeted, one-to-one video shot straight into the living room.
The measurement advantages of addressable, despite the limitations of household-level data, are significant too. Basically, you can tell exposed from control with high precision. The household maps to purchase data (behind a firewall), and presto, reliable attribution. As with online, the 1/1 targeting means measurement can be designed into campaign. Setting up exposed vs. control is straightforward (if not simple), and the results are reliable.
Broadcast has no knowledge of where ads land except via statistical samples. But stay tuned. Addressable protocols are being developed for over-the-air broadcast channels.
Ockham's Razor, a long-standing idea in philosophy, gives the story a boost. Ockham said, "What can be done with fewer [assumptions] is done in vain with more." Surrogacy of content for audience is a big fat assumption, and so is the surrogacy of age and gender for consumer need. With addressable TV, you find a home you like, and send an ad to it. Done.
So, as TV becomes more addressable, advertisers should try it on for size. You may discover it won’t fit very well, though, as part of a GRP-based TV plan. But it may fit like a glove as targeted reach in a digital video campaign.
Great piece Ted! Spot on.
I agree with Dave. Good report, Ted.
Great article by a super smart guy. Would love to hear Ted's perspective (or others') about what is required for TV platforms that enable precision targeting like addressable and OTT to take off from an advertiser perspective. Greater penetration? A consistent, horizontal view of audience size & characteristics across ad sellers? More centralized buying? Something else?
David, since you asked, the first thing that is needed is not only more nationwide coversge but more commercial inventory. At present, the addressable TV guys are limited to a few minutes of ads per hour and, I believe, many are spoken for by local cable spot sales. Second, is the lack of program by program targeting selectivity in many cases and third, is the reliance on set usage as a surrogate for viewing. It's a terrible surrogate. Finally, there is the matter of determining whether a given household is "in the market" for a product or service. In the absence of specific home by home information--- often not available--- many addressable TV buys simply profile households by location or other variables, assuming that any home that qualifies is to be sent targeted ads. This is not that much different from the much ridiculed TV "targeting" system where all 18-49s or 25-54s are assumed to be of equal value. Some are valid targets; some aren't.
A couple of slight adjustments here, responding both to Ed and David, and Thank you both very much for your kind comments.
Dish, and I assume AT&T, offer like 50 demo targeting variables (which are as accurate as axciom or experion HH information ... so now you can distinguish easily between different sorts of 18 to 49 year olds (Income, Ethnicity, Kids, geo, etc.). They also offer (and I assume AT&T has this as well) program type propensity targeting. So, people in the top decile for "News", for example, for about 30 program types.
Typical inventory is last minute of every half hour, so, 4 30 second spots per hour. Across 50 million HH, that would be 200 million impressions per hour ... many billions per day, which is nothing to sneeze at.
The surrogate for viewing is more than just the set being on ... it includes any STB operation as evidence of a person in the room, and the reach is not counted unless such evidence exists. (Ask Rentrak). Probably as accurate as depending on someone pressing a button on a Nielsen box, or self reporting in a diary ... and the sample sizes are huge, close to 1/3rd of all sets for Satellite, and census for at least some Cable addressble like CableVision.
To David's question, there are some pretty good tricks. Here is one. If you have a TV campaign targeted to programs (as a surrogate for demo), just buy the demo you want across the addressable network, but only target program propensities that are not covered by the network buy. So, you will get everyone you want except the ones likely to have watched the Network TV you bought. In effect, optimizing cost per reach point.
Another trick is to use it like Digital. Just take all the email names of your converters, hand them to you supplier, and they will develop a privacy compliant custom segment of STB-ID's which can be trafficked. That datset can also be modeled out for reach extention. I would expect any of the Addressable suppliers can do this stuff.
There is a lot more cool wonky stuff having to do with integration with Digital campaigns, and the media suppliers are quite famliar with it.
Ted, thanks for that reply. Regarding the set usage versus viewing issue, I doubt that the use of set-top-box activity aside from the set being on is a useful surrogate for determining that someone is viewing as this sort of activity is not of significant magnitude on a minute by minute basis. Also, we don't know who in the home is involved in such activity. The plain fact is that the average person in a TV home watches only half the time whan a set is in use. That creates a huge error margin for the STB concept where targeting actual consumers is concerned. STB data will consistently indicate that a TV show is reaching higher proportions of younger and affluent homes---which is taken by the system as a straight overlay on individual viewing---when, in the same households, young and affluent adults are, actually, light TV viewers for most shows. The reason for this is that younger/affluent homes contain more people, hence their sets are in use more often---but a given adult---the consumer, who is the real target for addressable TV----is less, not more likely, to be watching.
Regarding the availability of commercial inventory, a few minutes of time per hour---which is shared with local spot cable sellers, I believe---- poses a huge problem limiting the growth of addressable TV---even if it can deal with its other issues. Already we are hearing the addressable side trying to convince cable channels to allocate proportions of their in-program national ad time to addressable TV sales as a possible solution ---but this idea does not seem to be well received, currently. What the future holds remains in question and I wish the addressable TV folks well--as the basic concept---if deliverable at a reasonable cost and scale---certainly is an appealing one.