In what it claims is a TV audience targeting first, AT&T will enable advertisers to plan and buy TV shows based on the composition of wireless users who watch them. The data, which is one of two new audience segments being announced today as part of AT&T AdWorks’ “TV Blueprint” system, will enable brands to target shows based on how many viewers they have with specific wireless devices, operating systems and data usage plans.
The data, which comes from AT&T Wireless’ 70 million-subscriber database, will be fused with the TV audience data from AT&T U-verse subscribers, as well as other cable and satellite TV operators, to create a new composite view of TV viewing by wireless users. While the data is aggregate, anonymous and does not yet include any mobile behavior data, it could become an important new hook for advertisers and agencies looking to target TV viewers based on the devices they use.
The new data comes at a time when so-called “device-targeting” is emerging as an important potential breakthrough for targeting consumers across mobile, Web, TV and other platforms, and is playing an integral role in many of the attribution systems being developed to help the ad industry understand which platforms are delivering the best consumers for their campaigns.
While the new mobile data is obviously valuable for mobile and wireless industry product and service marketers, it could also play an important role for non-endemic brands for who see mobile usage as important signal. The data could segment users based on when their current wireless plans are expiring, for example, and could be used by carriers to target renewals or conversions. Ditto for hand-held devices and operating systems.
How that data would be used by non-endemic advertisers is unclear, but it conceivably could be used to retarget mobile users via TV, which would represent an ironic role reversal as a number of online data management platforms -- Collective, for example -- currently enable TV advertisers to retarget their viewers when they’re online.
AT&T AdWorks claims its Blueprint targeting system currently includes anonymous data from 5.9 million households via 15 million AT&T U-verse TV set-top devices.The second set of audience data being announced today enables advertisers to target TV shows based on individual members of the TV household. In the past, brands had to target TV shows based on whether a particular demographic was a member of the subscribing household, but now it can target explicit demographics individually.
Assuming that I understand the AT&T press release, it appears that the TV "Viewing "data comes from a smaller subset of set top boxes which provide only household usage data, not viewer data. As I have pointed out before, the fact that a TV set is tuned in is a very poor indicator of whether the targeted person in the home is the one who is watching. For example, across all dayparts, the chance that the person you want to reach is present when the TV set is on is about 50%. In younger and affluent homes with several children this figure drops to around 40% because there are so many more people residing in the home, plus visitors, who might be in attendance; in older homes where there are only one or two residents, the odds that the person you are targeting is watching are far higher---around 75%. As a result, if your stated target is women aged 25-54 with HH incomes of $100,000 or more, many of the TV shows that seem to be reaching this target, based on set usage, are reaching their husbands or a teenager or a child, and not necessarily the woman. Worse, the finer you slice and dice your targeting----going for ever more selective segments---the larger your error margin becomes. If one looks at Nielsen data the basic contradiction between set usage and viewing data is clear. According to the set usage findings homes headed by younger adults, larger Families and affluent households are well above par in their TV usage----considerably higher than older homes. When the viewer data is analyzed, the exact reverse applies----older persons are, by far, TV's heaviest viewers; younger adults and those with the highest incomes watch much less frequently.
Actually this is exactly what the new single person HH data in the AT&T AdWorks TV Blueprint product solves for. Using actual viewership data from single person households as opposed to data skewed by households with multiple people, marketers can be confident they are actually reaching their target audience and not simply the person who watches the most TV in the house.
We use this viewership data to then develop targeted media plans that are more likely to reach the advertiser's target audience in programs they are viewing.
Also, as point of clarification, the TV Blueprint product uses viewership data from all 15MM set-top boxes. The single person HH data comes from a smaller sub-segment of single person HH's.
Thanks for that clarification, Maria. Since single person households are not representative of all homes in terms of demographics----being mostly very old or, to a lesser extent, very young---I would then ask whether your single person households are representative of all households---and viewers. Or have you, somehow, been able to identify a single "respondent" in all homes, including those with more than one resident? Or, do you statistically "sample balance" your single person home findings in some manner to account for demographic imbalances relative to all homes? In other words, how representative is your TV usage sample? I'm not arguing----just asking.
Ed, you bring up some very good points. In utilizing set-top box viewership behavior to create more efficient data-optimized media plans to better reach specific audience targets, we wanted to develop an option that would allow our clients to get a “cleaner” read of TV viewership for individual attributes like age (especially narrow ranges) and gender that, with set-top box data, are often muddied by multiple people watching the TVs in a home. Limiting the set-top box data to single-person HHs for those targets achieves that by eliminating much of the multi-person clutter. At this stage, we have not gone down a nationally representative “panel” route although that is something we are exploring with AT&T Labs to identify what would be the best approach to address any potential skews that may exist in the census-level STB data. As with any methodology there are pros and cons to consider -- one of the primary benefits of TV Blueprint is its ability to provide viewership insight into longer tail/niche networks that are underutilized by advertisers because they aren’t measured by Nielsen’s panel. Maintaining that benefit (with a sufficiently large data set) is often at odds with traditional panel methodologies.
Thanks, again, Maria for your reply. I think you have approached this in a very sophisticated manner. As for solutions, it may be possible to use a combination of sample balancing and ascription----based on known audience comp ratios-----to generate more representative data, which could be "validated" by comparisons of topline findings with those of standard TV rating services. Anyway, good on your new initiative. Anything that helps advertisers to improve the targeting of their TV buys is a most welcome development.