Few cable subscribers realize it, but each time they switch channels, their TV provider makes a note of it. Today, that data is primarily used for internal research and to inform ratings. But newly published work from researchers at AT&T shows how it could also be used to make TV advertising more compelling.
The methods tested by researchers at AT&T Labs in Florham Park, New Jersey, are detailed in a paper that was presented at the International Conference on Data Mining last month. While they don’t allow for ads to be targeted to individual viewers, they could make TV commercials more effective—and more valuable.
TV advertising is today bought and sold in much the same way it has been for decades. A company that wants to advertise works out who it wants to see its message, and tells that to a TV station, where experts use ratings information from surveys, and their own judgment, to decide when best to run the commercial and how much to charge.
By contrast, few humans are involved in the process of targeting online advertising. Advertisers choose their target audience, and software does the rest, automatically matching ads with specific Web pages and even people (see “High Stakes in Internet Tracking”).
Data from the set-top boxes of an unspecified “major television provider” (presumably AT&T) was used in the new research. The data—anonymized to remove identifying data such as names—included details of channel changes and volume adjustments plus some demographic information. By assuming that someone was watching a channel any time a TV stayed tuned to it for more than 20 seconds but less than one and a half hours, the researchers built up a record of which subscribers had watched what channels, and when. That was used to predict when different demographic segments would be watching TV in the future—predictions that could be used to plan when to show commercials.
The AT&T researchers tested that approach with two large TV advertisers that they identify only as an “upscale car manufacturer” and a “large investment/retirement planning firm.” Both companies planned four-week-long campaigns using the conventional manual approach. The two campaigns were aired on schedules devised using the conventional method, but once they had aired, it was possible to use data from set-top boxes to show that the more-targeted system would have delivered more-targeted results. The big-data approach would have allowed a commercial to reach around 18 percent more targeted people for the same outlay on airtime. For a campaign willing to run up to six ads in any one hour, the cost of achieving 1,000 views from targeted people would be about half that of the conventional campaign.
Heather Way, an advertising technology analyst with Parks Associates, says that AT&T’s results underline that TV companies, advertisers, and viewers all have something to gain if the TV commercial business becomes more like online advertising. Says Way, “It’s more efficient for the ad buyer; the provider can get a higher rate for their ad; the consumer is not bombarded by a bunch of content that’s not relevant.”
There is hope within the industry, Way says, that one day such approaches could make it possible to schedule ads in real time in response to who is tuning in, and perhaps even target them to individuals.
Most people don’t know that their TV provider tracks channel changes. Some advocacy groups may protest the possible privacy implications of such techniques, but Way believes viewers will likely be even less concerned than they are about online ad targeting. “It’s a medium that consumers are just more comfortable with,” she says.
AT&T is not the only company exploring new uses for set-top-box data, says Way, although most efforts are focused on replacing the use of ratings based on surveys to guess at viewing figures. “They’re all looking to mine that set-top-box data and use it better and use it for audience profiles,” she says.
AT&T declined to respond to questions about its TV targeting research and so it is unknown whether this might become more than just a research project. But the company last year launched a new service called AdWorks that offers advertisers ways to measure TV campaigns using set-top-box data and to make TV ads interactive via a subscriber’s remote.
DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.
“This is a profound moment in the history of technology,” says Mustafa Suleyman.
What to know about this autumn’s covid vaccines
New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.
Human-plus-AI solutions mitigate security threats
With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure
Next slide, please: A brief history of the corporate presentation
From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.