Podcast ad revenue grew twice as fast as total online ad revenue in the US last year, but it still represents just a small segment of the country’s advertising spend. Perhaps that’s because, for some marketers, podcasts just seem too risky.
Advertising in the burgeoning space where many podcasters are known for putting their own spin on ad copy could seem like asking for trouble.
“Certain categories, like banking, for example, are very conservative with their messaging,” podcast company Acast’s global head of ad operations, Kevin McCaul, told Marketing Brew. “We have seen brands sit on the sidelines because they thought that podcasting was just the Wild West and there wasn’t control.”
That’s not good news for companies like Acast that generate revenue by selling podcast ads or anyone else trying to make money by podcasting. To address this hesitance, stakeholders from across the industry have been joining forces to develop brand-safety solutions for buying podcast ads.
Early network partnerships
Podcast ad buyers and sellers told Marketing Brew that clients in the space started seriously asking about brand safety as far back as two years ago. That’s no surprise given the pandemic and the political and social unrest that dominated the news cycle in 2020.
“More recently, over the last six, nine months, we’ve seen this idea of brand safety really, really become more of a challenge vocalized by our clients,” said Jennifer Laine, head of marketing, innovation, and special projects at audio ad agency Oxford Road.
Major podcast companies, of course, have large sales teams of people who are hyper-familiar with all the shows they sell ads for, but for more risk-averse advertisers, network sellers were looking for something additional “in our back pocket,” McCaul said.
To help achieve that, podcast companies have been inking deals with brand-safety partners in droves.
Last summer, Acast tapped Comscore as its brand-safety provider, leveraging its “natural language processing, machine learning, and AI” capabilities to enable advertisers to choose podcast episodes “that align with their brand preferences,” the company said in a press release announcing the partnership.
Break it down: Acast forwards podcast transcripts to Comscore, McCaul said, which in turn helps analyze them for keywords and tags them based on risk level in each of the Global Alliance for Responsible Media (GARM) brand-safety content categories.
- These 12 categories, which include “terrorism,” “debated sensitive social issue,” and “misinformation,” are used across digital platforms to help advertisers avoid the “monetization of harmful content online.”
- For instance, an episode that is specifically about gun violence or a shooting would be tagged as “high-risk” in the “arms and ammunition” category, McCaul explained, while a low-risk episode might just mention missiles in passing as part of a broader conversation about the space race.
- However, since GARM classifies “news features on the subject” as low risk, Acast would use AI to “also evaluate the context of the language and the frequency of the usage of the word,” added Elli Dimitroulakos, global head of ad innovation at Acast. For example, “if the episode refers to guns multiple times in a context around mass shootings, then it is considered high-risk,” she said in an email.
- Acast primarily uses this option for pre-produced ads to make sure they’re “inserted to the best-suited content,” McCaul said, providing that extra layer of AI-endorsed safety on top of the ad sales team’s knowledge of the content. For host-read campaigns, the company tends to rely more on its staff than technology to familiarize brands with the shows they’re working with—if buyers haven’t already vetted those pods themselves.
Earlier this year, iHeartMedia invested in audio intelligence platform Sounder, at the same time announcing intentions to tap into its AI and machine learning solution “to ensure advertisers can confidently invest” in its shows.
SXM Media—the advertising revenue org for SiriusXM, Pandora, and Stitcher—started working with Comscore via its AdsWizz arm in 2019 in a similar manner to Acast, according to Maria Breza, SXM Media’s VP of audience data operations and ad quality measurement.
Comscore can be leveraged “at the moment the ad is inserted,” Breza explained. She said that shows with host-read ads are usually handpicked by advertisers ahead of time, so while that does require brand-safety analysis, it’s not the focus for Comscore podcast partnerships.
“We see the most value for advertisers who have produced spots or who are buying announcer-read spots that will run across a wide selection of shows and episodes,” Breza said.
But Breza noted that SXM Media’s brand-safety approach is also not all about tech; its sales team works closely with buyers to help them understand the nuances of podcasts compared to the other kinds of content they buy.
Take the true-crime category, one of the most popular in podcasting: “It probably isn’t a brand-safety infraction for a brand to be next to ‘death and injury’ in that case,” Breza told us. “If you’re a brand and historically you have a low tolerance for death and injury, if you put that kind of anti-target against your podcasting buy and you’re trying to buy true crime, that’s going to be something that just doesn’t work.”
Emerging partnerships
Despite these types of partnerships becoming increasingly commonplace, brand-safety solutions for podcasting remain more nascent than for other mediums.
Get marketing news you'll actually want to read
Marketing Brew informs marketing pros of the latest on brand strategy, social media, and ad tech via our weekday newsletter, virtual events, marketing conferences, and digital guides.
“Advertisers aren’t yet entirely confident that their podcast ads will run in brand-safe environments because the development of brand-safety technology in the podcast space is still in its infancy,” said Eric John, VP of IAB’s Media Center, over email. “AI solutions that could provide brand safety at scale are available, but there is a lack of uniformly accessible and standardized podcast transcription data which has hampered widespread rollout of these solutions.”
For instance, IAB’s December 2020 guide to “brand safety and suitability in a contemporary media landscape” doesn’t include any mention of podcasts.
IAB’s Tech Lab and Media Center are facilitating conversations with podcast industry leaders to address this hurdle, according to John. The Tech Lab put together a “podcast-transcript clearinghouse working group” in Q2 of this year that discusses topics like how to create a more scalable solution for creating, hosting, and analyzing episode transcripts, John said.
The trade org isn’t the only party working on advancing podcast brand-safety tech. As recently as June, Spotify teamed up with digital ad-verification firm Integral Ad Science to create what the two companies called a “third-party brand-safety solution for podcast advertisers,” in a statement. This development came on top of Spotify’s existing brand-safety membership in GARM.
This summer, Lee Brown, Spotify’s VP and global head of advertising, told Marketing Brew that it’s working with advertisers to “adjust how they think about applying their legacy models of brand safety to a medium like podcasting, where there’s some nuances.”
Outside the networks
Others in the industry are finding uses for brand-safety tech outside of pre-produced ad campaigns, which still account for a smaller share of podcast ad revenue than host reads, according to IAB. For the 55% of podcast ads that are host-read, brands can be left with the question of how to vet hosts and their shows more holistically without spending countless hours listening to podcasts.
For those on the buy-side who might want to conduct their own analysis of content across multiple networks, agency land has also been developing podcast brand-safety tech tools of its own.
Podcast media-planning platform Magellan AI has developed a tool to “empower advertisers and agencies to review the content in shows based on their own criteria” instead of using the GARM categories for both host-read and pre-produced ads, co-founder Cameron Hendrix told us.
Clients can create what Magellan AI calls “keyword buckets,” then choose podcasts on its platform to analyze for those words. The Magellan AI tool then produces a report that shows when and where those keywords appear, which clients can analyze for additional context, according to Hendrix.
Magellan AI Airchecks screenshot
More recently, Magellan AI added an additional option to use the tool for “post-buy verification,” also referred to as “error checks,” Hendrix said. Advertisers can now also use their custom keyword lists to trigger alerts if any of their ads do end up running against content that could potentially cause concern.
Additionally, Oxford Road’s Laine said that the agency has been working with Barometer, a tech company focused on using AI to provide podcast content ratings to advertisers, on a brand-safety tool based on the GARM framework.
“It’s an AI-powered tool that contextualizes audio for sentiment or sentiment analysis and provides what we’re calling a ‘nutrition label,’” Laine said.
Barometer screenshot
One of the use cases for this tool is the potential to discover new inventory that could align with a brand’s “current risk profile,” Barometer CEO Tamara Zubatiy told us.
If a brand is advertising on 100 pods, a buyer could input them into the platform to run an analysis that might find, for instance, that 80 of those shows are classified as “low” or “no risk” of profanity.
“So it helps you discover what your risk profile is, and then you can make sure that the shows that you’re currently buying continue to reflect that,” Zubatiy said. “You can also then use that as an input set to filter inventory for discovery purposes,” or for post-campaign validation to ensure episodes where ads ran were compatible with the risk profile.
The Barometer tool sorts episodes into the GARM categories based on risk, but can also rate “tolerance” for certain topics, Laine explained. For instance, if a host shares an opinion on a sensitive social issue but does it in a way that’s respectful and acknowledges an opposing view, Barometer would rate the episode as “high tolerance.”
The tool highlights these instances in the time-stamped transcripts of the episodes so advertisers can fast forward to them, listen, and then decide for themselves if the content aligns with their brand.
This could allow podcast advertisers to potentially avoid some of the issues other brands have faced in trying to avoid risky content, like shifting ad revenue away from legitimate news outlets in an attempt to avoid certain topics.
“Some brands just want the absolute assurance that they’ll never be involved in controversy at all, which nobody could actually confirm,” Laine said. “You never know what’s going to happen.”
But ultimately, she added, these brand-safety tools could help the podcast industry demonstrate to major advertisers that podcasting is a safe and legitimate medium, giving them a reason to test it out.