Ad fraud has come to audio, according to report

Audio, largely considered an up-and-comer compared to ad channels like social and CTV, is showing signs of maturing. Some signs, like podcast ad revenue crossing the $1 billion milestone in the US, are positive. Others, like potential ad fraud schemes, are less so.

Last month, brand-safety firm DoubleVerify said it uncovered a “scheme” it’s calling BeatSting involving alleged audio fraud that has cost “unprotected advertisers” up to $1 million per month, or potentially as much as $20 million over the past few years, according to Roy Rosenfeld, head of the company’s Fraud Lab.

“You’re seeing an explosion in podcasts and other types of audio inventory, but it still doesn’t really meet the increasing demand,” Rosenfeld told Marketing Brew. “As soon as you have this imbalance between demand and supply, especially in these high-engagement channels like CTV and obviously audio as well, this just creates an opportunity for fraud and invalid traffic.”

DoubleVerify isn’t the only brand-safety firm eyeing the topic: Last summer, Integral Ad Science said 87% of media experts it surveyed were “concerned about ad fraud in audio.” A month later, IAS announced that its ad measurement tool would be available for use on Pandora, enabling advertisers to detect invalid traffic on the platform.

Still, some marketers said they haven’t identified it as a major concern quite yet, at least among podcast advertising, partly because they’re not purchasing ads programmatically at a large scale.

Gotta catch 'em all

DoubleVerify first caught wind of BeatSting in 2021, according to the report, which said the alleged fraudulent activity came in the form of more than 60 mobile apps, including jazz and rock music players, as well as an app geared toward anime and manga fans.

The apps are real and downloadable, Rosenfeld said, but claimed that the impressions “never actually occurred on the apps.” Bad actors can “essentially make up the fact that there’s an audio stream that requires an ad to be stitched in,” Rosenfeld said.

They do so by creating traffic that mimics the way real users might listen, triggering a request for an ad exchange or server to insert an ad, Rosenfeld explained.

Because the streams that lead to the ad request aren’t real, DoubleVerify can use AI models to detect when activity on an app doesn’t match the way a person might actually engage with it, according to Rosenfeld.

DoubleVerify didn’t break out how much of the activity it detected took place on specific audio channels, including podcasts.

Small potatoes?

Rosenfeld said he expects this activity will continue. But some audio buyers don’t seem to be worried, at least when it comes to podcast ads.

Perhaps that’s because, compared to fraud-related losses in other channels, $1 million per month is miniscule. Jordan Bentley, CEO of audio DSP Audiohook, explained that “audio historically has been less of a target due to lower scale relative to display and video.”

“[Fraud] is probably…more prevalent in programmatic,” Rosenfeld said, and programmatic sales make up just about 2% of total podcast ad revenue, per the IAB. (Though it’s worth noting that podcasting accounts for just one portion of digital audio ad revenue.)

That could explain why some agencies told us that it hasn’t been a problem. Michele Madaris, media director at independent marketing agency Boathouse, said she hasn’t encountered it in audio, noting that her team does the majority of its audio buying directly as opposed to programmatically, but wasn’t surprised to hear it’s becoming an issue.

Similarly, Lisa Jacobs, VP of media operations at Ad Results Media, said that ad fraud “hasn’t been an issue” for the radio and podcast agency. That’s largely because it focuses on host-read endorsements and performance marketing, she explained. “As we expand our programmatic buying, I’m sure this will become more of a topic. But…I don’t see large risk here for our clients.”

Measure up

Others told us they have seen what they consider to be fraudulent activity, in part because attribution services they work with have flagged it. Miranda Romano, VP of media operations at performance audio agency Oxford Road, said her team works with podcast attribution services like Podscribe, which has helped identify potential fraud.

“Especially since we started using pixels to track performance with companies like Podscribe, we are definitely seeing a higher volume of it on podcasts,” she said. “We also track streaming [music], and we don’t see as much suspicious activity as we do in podcasts.”

Grant Durando, director at marketing agency Right Side Up, said he has noticed “some isolated areas of ad misbehavior” thanks to attribution partners like Podscribe and Magellan.AI.

Ultimately, Durando said he’s “skeptical that there’s a significantly scaled, deliberately nefarious ad fraud scheme” going on with regard to podcasts.

“Overall, any ad fraud scheme that currently exists or is in the process of being conceived will be subverted by a few simple, common-sense moves that we as an industry are making,” such as using third-party pixels to track ad delivery, he said.