New AI tool can help advertisers vet podcast hosts based on news they make outside of their shows

This just in: Podcast hosts can be controversial. Maybe you’ve heard of Joe Rogan?

More breaking news: Brands like to pay podcasters to endorse their products and they’re often willing to give up some control over ad copy in the hopes that an authentic anecdote will spur a sale. But, in an even more shocking update, advertisers also care about this thing called “brand safety.”

There are already plenty of tools at their disposal to help them avoid running ads on podcasts or episodes they’re uncomfortable with. But hosts sometimes make news outside of their podcasts that can make marketers squirm.

That’s why Barometer, an AI-powered platform that rates podcasts based on “brand suitability,” teamed up with audio ad agency Oxford Road to create a tool called Host Intelligence, which analyzes news and online content surrounding podcast hosts with the goal of helping advertisers better assess sponsorships.

Hi, I’m the problem

Brands and agencies typically “manually” conduct audits of the podcast hosts they’re considering sponsoring or already work with, which can mean “hours and hours” of Googling and reading articles, according to Jennifer Laine, Oxford Road’s head of marketing and special projects.

As an agency, “it’s virtually impossible” to conduct these audits across the many podcasts its clients advertise on, Laine said. Still, “so much of the host controversy actually happens outside the episode.”

“Even recently, we had a major host that a lot of our clients were sponsoring who was convicted of tax fraud, and of course, [they] were not talking about it in their episode,” Laine told Marketing Brew. An AI tool only analyzing episode transcripts wouldn’t turn up that news, “but for the brands who were sponsoring that [show], it was really important and relevant information.”

Sarah Marini, senior manager of audio partnerships at Athletic Greens, wrote in an email that even if the content of a podcast is considered brand safe, a host might not be, which could “be a reason to pass on an opportunity.”

While brands are increasingly investing in podcast ads, Eric John, VP of the IAB’s Media Center, said in an email that audio-focused IAB members have indicated “that brands need to have greater trust in their ability to identify brand safe and suitable content” before they fully take advantage of podcasts at scale.

In addition to episode analysis, “it is also critical for advertisers to have a line of sight on consumer sentiment about hosts and potential positive or negative perceptions of them in the news media,” he said.

Hi, I’m the solution

Host Intelligence became available to Barometer subscribers in September, and the company spent the past few months collecting feedback from users and plans to roll out an updated version by the end of December, according to co-founder and CEO Tamara Zubatiy. Four of the brands that tested it were among Magellan.ai’s list of the top 10 podcast ad spenders in August.

Before Host Intelligence, brands and agencies could use Barometer to conduct “contextual analysis” of podcasts they were advertising in or considering working with, Zubatiy said, based on the AI’s ranking of shows as high, medium, low, or no risk across GARM’s categories, which include “misinformation” and “debated sensitive social issue.”

Host Intelligence layers on top Barometer’s existing transcript-level analysis by analyzing 50,000 different content sources every day, tracking mentions of 2,083 podcast hosts, and weighting those mentions based on how prominent they are in the content, Zubatiy explained.

That news can include stories from major outlets, small publications, or “bot-generated articles” pulled from social media, Zubatiy said. There’s no guarantee that all social media content will be picked up, Zubatiy acknowledged, but “if it gets mentioned enough times, then it’s likely that we’re going to hear about it.”

Now, advertisers can not only see Barometer’s content risk scores of shows they track but also a bar on the top of their screens that tells them if the system found that the hosts of those shows have been in the news in the past week or the past month and whether the overall sentiment of that coverage is rated positive or negative by the tech, Zubatiy said. From there, users can click out to view the articles themselves.

The updated version rolling out in the coming weeks will also include news analysis dating back a full year, Zubatiy said, as well as a line chart showing advertisers how the AI’s rating of the sentiment has changed over that time.