Dynamic ad insertion isn’t the silver bullet for advertising. Could it get a boost with AI?

If you were one of the tens of millions of Americans who tuned into the Super Bowl this year, chances are you saw pop star/American treasure/master marketer Beyoncé teasing her new country album, Cowboy Carter, via Verizon commercial. That’s because the ad was “burned-in,” which means that it was pre-embedded during breaks in the game for all viewers to see.

But there’s another way that ads could technically appear in major television events, one that is sometimes touted as the future of the advertising industry. Instead of showing the same ads to every viewer, ads during the game could be personalized for different viewers through dynamic ad insertion, meaning that viewers would be shown different ads based on their attributes—which means that while some viewers could have been served the Beyoncé ad, others might have seen a different ad entirely.

In theory, dynamic ad insertion, sometimes referred to as DAI, could lead to increased personalization and engagement, which could result in higher viewer (and customer) retention, Dan Rayburn, conference chairman of the Streaming Summit at NAB Show, told Marketing Brew. In practice, though, it has left something to be desired, Rayburn said: “It doesn’t scale, and it doesn’t work well.”

Despite those challenges, experts told us that the technology could still be promising—as long as some of the wrinkles get ironed out.

Currently, in a traditional linear environment, “burned-in ads are the only way to go,” IAB Tech Lab CEO Anthony Katsur told Marketing Brew. But “as the industry continues over the next 10 years to move toward a world where television is largely streamed, I would see less and less of a need for burned-in ads,” he said.

Keeping it dynamic

Server-side ad technology can show personalized ads to different consumers watching streaming content, like live programming, and amid a broader shift toward streaming, there is a “marked shift” toward the technology, especially from networks, MVPD providers, and cable providers, Katsur said. (In addition to streaming, it can also be used in the context of programmatic linear TV, he noted, but given the streaming boom, he doesn’t expect it to be widely adopted.)

While useful for targeting specific consumers, being able to deliver different ads simultaneously during a big broadcast like the Super Bowl can risk a broken spot in the programming, according to Katsur. As of now, burned-in ads are “much more reliable than dynamic insertion,” he said.

There is some hope that generative AI tools could potentially be used in the future to help deliver personalized creative efficiently—and, in some cases, create those personalized ads, too, Katsur said. He used the example of a hypothetical McDonald’s ad campaign to demonstrate how AI could convert user data into targeted ad creative: One viewer might see an ad promoting a Big Mac, while another might get an ad featuring french fries.

“The interesting thing about generative AI is that it could, to an extent, dynamically stitch those assets together on the fly, based on the understanding of a [consumer’s] predilections toward certain McDonald’s foods,” Katsur said.

Get your facts right

Personalizing ads at that scale can pose another challenge for media companies themselves: sifting through the creative to approve (and, in some cases, not approve) before airing, Dave Morgan, founder and executive chairman of TV advertising platform Simulmedia, told Marketing Brew.

“There’s an incredible approval process in something like the Super Bowl, but there’s also a process in the local TV station in Des Moines, Iowa, or Erie, Pennsylvania,” Morgan told us.

Again, generative AI could come in handy. In a lot of markets, humans manually check dozens of ads each day, but not, say, tens of thousands, Morgan said. AI could be used to check large volumes of creative and identify ads that might not meet standards, Morgan said. That technology is already being deployed, like at Amazon’s AWS, where there is AI-powered ad verification, he said.

Throw a curveball

But there’s one thing that DAI can’t really solve: The fundamental unpredictability that comes with live events, as well as streaming’s historic challenges handling potential surges in viewership, Katsur told us.

“If it’s a boring Super Bowl, if it’s a blowout by the third quarter, people will tend to tune out, so you will lose viewers,” he explained. “However if it’s a shootout or a really close game, [like] the last Super Bowl...you’re going to see viewership potentially spike,” Katsur said.

To make the streaming experience more reliable, and to help make DAI a reality in the streaming space, the right infrastructure and server support will be crucial, Katsur said.

“There’s still some work to be done from an infrastructure perspective for streaming to be as reliable as cable and satellite have been for decades,” he said.