A GroupM and iSpot study spotlighted the problem last year, but a permanent fix remains elusive

Buyers remain worried that they’re paying for ads that are running on TVs that have been turned off, an enduring frustration that is caused by technology issues or, some say, a lack of transparency.

In June 2022, GroupM and measurement firm iSpot found that 8% to 10% of impressions on connected TV ran when the TV was off, costing advertisers about $1 billion a year in wasted dollars, the Wall Street Journal reported at the time.

The report diagnosed one cause as from external devices delivering streaming content—a user could turn off the TV device without the streaming device knowing and could still send bid requests as a result. Due to the error, 17% of impressions coming via external devices were delivered when the TV was shut off, the report found, while the problem was virtually nonexistent on smart TV apps.

Over a year later, buyers are still scrutinizing their ad buys to make sure ads don’t run when the TV is off. While two buyers say the problem has improved, it still persists.

“[Publishers are] rigging the game in their favor, and that’s why we need full transparency,” said one buyer, who requested anonymity to freely discuss industry relations.

Advertisers’ struggle to make sure their ads appear on a TV that’s actually playing is emblematic of the challenges of buying streaming video, where the ad tech and deal structures are still designed for linear television or online video rather than connected television, a more hybrid medium.

But transparency is easier said than done when the industry is not fully in agreement with the root cause of the problem. Likewise, measuring is not straightforward while the industry is still developing tools and standards.

Are you still watching?
Buyers are working on fixing the problem, though they’re not always unified on its source.

MediaMonks started evaluating the rate of ads running with the TV off a year and a half ago, said head of brand investment Matthew Kramer, whose findings were similar to GroupM’s figures of between 8% and 10%. Now, instances of TV off inventory are in the low single digits.

Kramer attributes this decline to the agency prioritizing more direct deals with content owners.

“It’s not necessarily the publisher’s fault because there are so many touchpoints that they might not know themselves,” he said. More hops in the chain mean more opportunities for miscommunication or more opportunities for fraudsters to intervene.

Though Kramer said fraud might cause impressions to fire on an otherwise off TV, a second buyer requesting anonymity who also tracks the problem said that the root of the issue is the technological challenges of different devices communicating with each other. As such, doing direct deals with publishers wouldn’t necessarily help.

That buyer said that the rate of ads airing with the TV off from external devices has decreased markedly from last year, though wouldn’t share specifics.

These results came after the buyer’s agency worked with publishers to present “are you still watching” labels to viewers after three or four hours of viewing.

Two other buyers blamed the problem partially on publishers seeking to pad their bottom lines with wasted impressions.

Whatever is to blame, some industry insiders aren’t sure how long TV-off ads will plague the industry. Smart TVs, where ads airing with the TV off are less of a problem, continue to gain market share against external devices. Viewing time on Smart TVs grew 31% between the second quarter of 2022 and the second quarter of 2021, while connected TV device viewing only increased by 7.8%, according to streaming media intelligence firm Conviva.

“I think this problem will phase itself out,” said Shailin Dhar, founder of digital advertising measurement firm Method Media Intelligence.

A lack of data points
Buyers keen to know how much they are wasting on ads airing when the TV is off still have trouble finding one single source of truth.

MediaMonks deduced TV off rates by comparing automatic content recognition data from smart TVs, data that shows what is playing on a smart TV and data from set boxes.

“There doesn’t seem to be one silver bullet,” Kramer said. “You have to look at several different data points. Right now, you have to do investigative work.”