Fine-tuning TV ratings
TVision tracks how glued viewers are to shows, ads
Nielsen ratings remain the gold standard for TV networks and the advertisers trying to influence viewers, but a show or ad’s impact is unclear in a world where many have one eye on the TV and the other on their phone.
A startup born out of Massachusetts Institute of Technology is aiming to do just that, by using cameras and artificial intelligence to detect how closely people are paying attention to the shows they’re watching.
“We are really good at understanding audience attention,” said Yan Liu, chief executive and co-founder of TVision Insights. “We use deep learning, machine learning, essentially computer vision to understand what happens in front of the screen.”
Using cameras in about 2,000 homes in the U.S. — all people who have agreed to take part and are compensated — TVision tracks whether viewers’ eyes are glued to the screen, paying more attention to a phone or computer than the show, or even out of the room completely.
“We can tell when you’re paying attention and when you’re looking down at your iPhone,” Liu said. “If a company can find a high-attention show for the same price, that’s a really good opportunity.”
TVision’s system was able to determine, for example, that viewers often pay more attention to commercials than the actual game during the Super Bowl. A Bud Light commercial featuring a knight attracted the most attention, TVision said. TVision’s data can also help companies track when a marketing campaign is getting stale.
“If people get bored of it, they don’t want to watch it anymore,” Liu said. “If we understand nobody is paying attention, maybe it’s time for new creative.”
The company works with major broadcasters including ABC and Turner Broadcasting, and advertisers including Mars and Nestle.
Still, TVision is not aiming to topple Nielsen, Liu said. Instead, the company is trying to position itself as complementary.
TVision is also in about 1,000 homes in Japan and will be expanding to the U.K. soon. Liu said personal information is never sent from a person’s home, including their image.
“We work really hard to protect people’s information,” Liu said.
TVision, founded while Liu was in the middle of completing his MBA at MIT, has raised close to $10 million in investor funding but is preparing to close another round of funding — likely between $10 million and $15 million — soon, Liu said. He said the company’s revenue grew by 450 percent last year.