How podcasters are talking about the Israel-Gaza war

I have been paying close attention to Brookings’ political podcast database over the past week and a half to see how discussion of the war is taking shape. This is a highly underrated tool and an essential one if you care about the state of political podcasting. It looks at the top political podcasts on the left and right on Apple Podcasts (i.e., no Rogan) and catalogs what they’re talking about. One feature breaks out the most-discussed topics on each side, and the language used is telling about how conservative podcasters are approaching the topic versus liberal ones.

Immediately after the terrorist attacks on October 7th, conservative podcasters cornered the conversation. While “Israel” was the number one topic for conservative shows, it didn’t crack the top 10 on liberal shows. That has changed as the conflict has escalated into war, with “Israel” being the number one topic for both sides in the past week, but you can still see the difference in language and approach.

For conservative podcasts, you can see the hawkish bent, with “Hamas,” “Biden,” and “war” toward the top. On the liberal side, you see “Gaza” and “Jordan,” likely focusing on the humanitarian and diplomatic aspects. You can also see an avoidance of topics they would rather not touch. “Hamas” doesn’t appear on the top 10 for left-leaning podcasts, while “Gaza” ranks 10th on right-wing shows.

By looking at the language these podcasters are using, we can see how the narrative is being shaped. This is especially true when you take this information in conjunction with the dashboard’s count of conservative podcast episodes versus liberal ones (or moderate, but there are very few of those). Among prominent political podcasts, conservative shows outnumber liberal shows by three to one. For now, in podcasting, the conservative perspective on the war, and the language they are using, is dominant.

– Ariel

So while everyone else is flailing, is Patreon thriving? Or trying to? The creator platform, which monetizes some big-name podcasts like You’re Wrong About and Chapo Trap House, has acquired ticketed digital events platform Moment (Patreon spokesperson Alexandra Duff declined to disclose the value of the deal). Patreon will integrate Moment’s features, which include ticketing, livestreaming, and merch sales, over the course of the next few months.

As my Verge colleague Mia Sato wrote about a few weeks ago, Patreon has ambitions to become more than just a paywall. With a new Discord-esque messaging system and amped-up creator profiles, Patreon is becoming more like a social network at a time when other social networks (cough, X / Twitter / the fifth circle of hell) are imploding or moving toward TikTok-like recommendation media. Having a live events capability builds on that effort, giving fans something to watch and talk about in the same place.

– Ariel

Earlier this week, Hot Pod ran the first part of an interview with Spotify’s director of global ad platform integrity, Dave Byrne. Now we have the second part, going deeper into how Spotify’s ad safety tools work, just for Insider subscribers.

One sentiment that was frequently brought up during Podcast Movement was that advertisers have become even more picky and sensitive about brand safety. Do you believe that is true?

I don’t think that advertisers are necessarily more picky. I think that what’s happened over the last while is that advertisers are more aware of what’s actually out there and what’s available to them. The brand safety movement has resulted in an increase in transparency. That old adage comes in, you know — “50 percent of my advertising doesn’t work, but I don’t know which 50 percent.”

Before, you could almost say, “I don’t know where 50 percent of my ads show up” — like who they show up next to, what kind of content they show up on. That’s becoming more transparent as brand safety continues to be something that people invest in. As a result of that, we’ve got greater level of controls and greater level of functionality. I think what’s happening now — rather than brands being more picky — they’re like, oh, this is the type of content that is really able to reflect their values and principles.

If we think about the breadth of podcasts that are out there, there’s truly so many different niches that advertisers can reach. They can really hone in on the right audience with the right creators and the right messages that are being shared. And [those opportunities] are only going to grow — especially in the podcasting space.

Can you give me an idea of what kind of information an advertiser receives from Spotify on individual podcasts?

One of the things I’ve noticed, coming from other platforms to Spotify, is that we’re still very nascent in regards to the knowledge that’s out there about the digital audio space.

For a lot of our advertising partners and a lot of our agency partners, we’re still in this phase of doing one-on-one brand safety, like explaining how we are investing in our people, our policies, our technology to actually create a brand-safe ecosystem.

We’ve been working with [Integral Ad Science], which, you may be aware of, are a big brand safety provider across display and video. We actually did a piece of research with them, where we analyzed 12 months of podcast episodes that we had monetized to verify that this content was over 99 percent brand-safe. That’s one of the things we show to our advertisers and agencies, to show that we’re working from a very safe and secure foundation.

We’ve also been investing in our first-party controls, features, and reports. Over time, we’re going to be building more first-party solutions into Spotify Ad Studio so advertisers can get even more there. One of the things we’re working with IAS on is post-campaign reporting, so a report showing not just how brand-safe the ad campaign was — but also how brand-suitable it was.

We’re still very much in a “build” phase right now. It’s about education. It’s about showing how we’re creating a brand-safe environment and continuing to build these post-campaign reports.

Can you talk more about how Spotify’s brand safety technology works? Is there a certain score that is generated?

We leverage technology to transcribe podcasts that get uploaded to Spotify. We take a look [at those transcripts] to get an idea of what’s being said. This technology has been trained on a vast amount of podcast episodes that have already been shared but also through our human expertise. That helps us categorize something firstly as being, “Hey, is this something that is fun and safe — or is this not suitable for monetization?”

We’ve trained this technology on hundreds of thousands of podcasts that we’ve already seen — but we keep humans in the loop to make sure that we’re making the most important decisions to help categorize those podcasts. It helps us then create targeting controls, like Sensitive Topics, and category targeting and episode topics.

How does [Spotify’s brand safety] technology take into account things like sarcasm and changes in vocal tone, for example, that may alter the meaning of a sentence?

A lot of this technology is proprietary, so I can’t go into much detail. But again, it’s a mixture of using our historical data to train our machines, to train our models — but also making sure that humans are in the loop to make sure decisions are being made in the right way.

– Amrita

That’s all for today. Amrita will be back with you tomorrow with a packed issue.