690. It’s Show Business, Baby!

How the podcast industry works

0:00Hey, I'm so stupid. You think? Oh, come on. All right. Let's hear about that. Oh, man.

0:12Tell me when you want me to stop trolling. You know, I'm a natural born bully. I've come to see that. I believe in justice. I do bullying for justice. This person, this bully for justice, is my friend Jenna Weiss-Berman. She is a podcast producer. We'll get to her very strong feelings about AI later. What's this episode about? It's a podcast about podcasts. We get about 10 to 20 emails a year of people saying, I listen to your show, and I don't understand how it works. It sounds like you all work kind of hard,

0:46but like, how do you get paid? Do you get paid? No, you just do it for free. It's so sweet. For the record, no, we don't make this show for free. So how does it all work? Let's pull back for a second. The entire podcast industry grosses maybe $9 billion a year globally. As industries go, that is relatively tiny. It's smaller than the bowling industry. It's smaller than the U.S. pet grooming and boarding industry. It is much smaller than the U.S. funeral home industry.

1:16But we're also talking about a relatively new industry. 2004 is generally considered the year that podcasting started. As of today, there are an estimated 620 million global listeners. For some people, it replaces talk radio or TV. For others, it opens new windows onto totally new worlds and new kinds of conversations. And for some listeners, a podcast is a place to ask questions. For instance, a Freakonomics Radio listener named Jerry recently sent us this email.

1:49There often seems to be a conceit that podcast hosts are scrappy individuals who are running the podcast from their home, she writes. But recently, I've been paying more attention to the long list of production credits that are acknowledged at the end. Who is funding all these people? Jerry had a lot of other questions about podcasts. Maybe you do too. So today, on Freakonomics Radio, a producer, an agent, an economist, a media executive, and an AI evangelist walk into a recording studio and tell us everything they know about podcasting.

2:21Maybe we should start with some etymology. Emerges iPod, which you can only buy in porn stores now, with broadcast, which it isn't. So both words that emerge to make podcasts make zero sense to anyone. Do you have a better word?

2:38Until there is a better word for it, this podcast starts now.

2:55This is Freakonomics Radio, the podcast that explores the hidden side of everything, with your host, Stephen Dubner.

Producing hit shows with Jenna Weiss-Berman

3:05I met Jenna Weiss-Berman a few years ago at a party for, big surprise here, people who make podcasts. So what exactly does she do? I like to help people figure out what their podcasts are going to be. I'm a pod doc. Oh my gosh. Cut that. That word is worse than podcast. No, do you know what's worse? A podcast doula. I could have said that.

3:36You just did. And we have it. Perfect. That's what I do. I help people bring their podcasts in the world. These days, Weiss-Berman runs a production company called Paper Kite Audio, an offshoot of Amy Poehler's Paper Kite Productions. And she's executive producer of the hit podcast, Good Hang, with Amy Poehler. So what makes a podcast host a good host? I think a person has to be very curious, a good listener. They have to be really passionate about doing a podcast.

4:08Amy Poehler was always going to be the best at this. Like, I knew that the day I met her. Do you need to make a good podcast to have a financially successful podcast? Yes, you do. And that does not mean that you need a famous person at the head of it. Sometimes it helps. Says the person whose podcast won a Golden Globe. I don't know about award shows, but when they get it right, it makes sense. No, that's not why it won a Golden Globe. It's because she's so great at it. And she's so committed to it.

4:39I want to say that I know I am new to this game. I have great respect for this form. I have great respect for all the people that I am nominated with. I am big fans of all of you, except for NPR.

4:53Just a bunch of celebs phoning it in. So, try harder.

4:59Yes, a celebrity might bring an audience one time. They will not keep an audience. There was a period where there were a lot of celebrity-hosted podcasts being launched. Obama's, Harry and Meghan, and many, many, many, many others. I think some people hear about the gold rush of podcasting. And they think that they can do it because they're famous and they can jump in and get an audience. And a lot of them come into it with no ideas. And they're sort of hoping that the people around them will come up with ideas.

5:33But hosts themselves have to be invested. Really invested. Like, really excited. Amy and I talked for two years about her show before we pitched anything. No, you're kidding. No. Tell me about that conversation. Well, she listens to absolutely everything. She understands what is really goofy about podcasts. What's funny that isn't maybe intentionally funny sometimes. And what is funny that isn't intentionally funny? Just some of the self-helpiness of podcasts.

6:03Not yours. Yours is actually helpful. We don't help anyone. Well, actually, that was kind of a goal of Good Hang, which is, like, we're not trying to change you. This is not, like, a self-betterment space. This is just a place that you can, like, have some fun. We love the audience and we just want them to be happy and be able to escape from the horrors of the world. People think of these things as very effortless. As you know, they're not. Good Hang takes a ton of work. I, like, really meticulously edit every episode. I go to every recording.

6:34So if you hear someone giggling in the background, it's me. You know, I do a lot of prep stuff. I deal with our advertisers and we do live events. So how much do you know and care about whether a podcast will be successful when you start working on an idea? I always know. Okay. So how do you know? Because that sounds made up. I know, but I do. I just, it's the only thing I know how to do. Have you ever made an unsuccessful podcast? Yeah, totally. Oh, well, then you didn't know. No, I didn't. Sometimes I've made podcasts that I was hired to make that I didn't think were going to be hits.

7:08I still had to make them. How does one go from a non-podcast young person graduating from college to becoming the podcast? What was the word you used? Doula? Was it queen? Was it doctor? I just make podcasts. I'm a producer. What was your first real job in audio? First real job in audio was an internship at StoryCorps. It's essentially like an oral history project that is affiliated with NPR and the Library of Congress. I would go to an internship there in the mornings from like 9 to 12.

7:39And then I was a collector at a corporate law firm from like 12 to 9 p.m. basically. Were you good at that? I was so good at it. I loved making people give money that they owed. The checks would be addressed to me. I would get like a $4 million check from the biggest corporation. It was absolutely exhilarating. What was it about audio or radio that made you want to keep coming to that? Were you an NPR kid? Yeah, I was. Like Western mass NPR kid. But I just liked audio storytelling more than anything else.

8:12Why? Because you could use your imagination. You can build a world. I just said storytelling and world building, which are two other terms I hate so much. But whatever. I feel like all good audio is like a good story, right? It doesn't matter to me if it's a chatty show, if it's a little comedy thing, if it's like a very complicated narrative series like I used to make with my company Pineapple Street that we would spend like a year and a half crafting eight gorgeous episodes.

8:44All people want to hear is a good, well-told, well-edited story.

Building Pineapple Street Media

8:52Weiss Berman founded Pineapple Street Media in 2016 with fellow producer Max Linsky. At the time, she was running audio at BuzzFeed. Max and I just kept hearing from interesting people who were like, we'd love to have your help making a show. I think Gimlet had just launched. There were a couple of networks at the time, but there was nobody who was doing like high-level production with partners. We didn't necessarily need to own it. We just wanted to help people.

9:22Mostly we wanted to help people who were already operating at the highest level of what they did. So we started with the New York Times, you know, they do what they do pretty well. They didn't have podcasts yet. Well, they had had podcasts and they kind of abandoned them because I started my podcast while I was at the New York Times. You did? Freakonomics? Yeah. At the time, the Freakonomics blog was on NYTimes.com. They just licensed that from us. And so I started the podcast and the Times said, well, we can distribute it.

9:54And so we said, great. And then it was successful. And then I said, hey, do you want to go into business on this? And they said, you know, we tried podcasts for a few years and it really didn't work. This is so many places. We tried podcasts for a few years and it didn't work. But everyone tried them so half-assedly. Can I not say that on the radio? No, I'm only laughing because you pronounced it so beautifully. Half-assedly. Half-assedly.

10:20Podcasting for absolutely everyone. For all media companies for so long was a thing that companies did not invest in or take seriously at all. For Pineapple Street, we sort of wanted to like level that up where it's like you have all these amazing journalists. Podcasting is becoming a thing. We can help you make good things. It doesn't just have to be something that you're like passionate journalists are doing in a closet in their free time. That's also how so many of the best podcasts got started.

10:51Reply All became a massive hit and was started as a side project of some people at WNYC. WNYC didn't want it. It went to Gimlet and it's probably the reason that Gimlet sold for so much money is that show. How long was Pineapple Street around? And tell me about The Exit. I think we probably sold it a little early. Yes, you're giving me a big nod. I'm nodding in sympathy, if not in agreement. Weiss, Berman, and Linsky sold Pineapple Street in 2019 to a company called Entercom, which was mostly known for its AM and FM radio stations.

11:25Today, that company is called Odyssey. The sale price was $18 million. That same year, Spotify decided that podcasting was a big deal and they bought several production companies, The Ringer and Gimlet and Parcast. Gimlet alone went for about $200 million. Where do you put us right now in 2026 on the arc of podcasting as a medium and as a business? Are we like early golden age? Are we way past golden age? I think this is the golden age.

11:57So many people thought that podcasting had its golden age whenever it was, like eight years ago when you were seeing all these companies selling for hundreds of millions of dollars. Wild. But that was really all speculative money and it was venture money. Some of the places that were selling were highly profitable and others were very early in trying to figure out their business models. It was a very frothy time, but it wasn't real money. It wasn't ad dollars. Like the ad dollars really weren't there yet. So what was it was just getting market share so that when the ad market or some other market matured that they'd be there?

12:31Yeah. So podcast listenership was growing rapidly, but podcast ad dollars were not growing rapidly. There would be these networks with 200 shows and they were supposed to sell ads on them, but then they would forget about 100 of them. And then there were also on the network sides, there were these huge minimum guarantees is what we call them, which is essentially an ad advance that's structured kind of like a book advance where you get a guaranteed amount of money from a network and then they have to find the advertisers to actually get that money.

13:04A lot of those deals started off really crazy, like where it would be a show that didn't yet exist with somebody who had some level of fame and suddenly they were getting, you know, $8 million a year to make something that they hadn't even decided what it was yet. And that really screwed these networks because they didn't have the download numbers to bring in the advertisers that they needed. So a lot of shows got shut down. A lot of people were fired. Yeah, I would say a lot of them, you know, had a great time for a few years and then they were restructured in a way that didn't make them happy, but some of them stuck with it, understanding the actual economics.

13:45What share of podcasts would you consider profitable? I'm including every guy in their parents' basement. Sure. Why is it always only the guys? Do women never live in their parents' basement? No, I'm just being honest. Women do not live in their parents' basement. I don't know, probably like 1% are profitable. This is probably a good time to say that one of the many quirks of the podcast business is that there is no industry-wide tracking system. So firm numbers are hard to come by.

14:15A lot of people do cite the same figure that Weiss Berman cited, that only about 1% of shows are profitable. If that's the case, can we really be in the golden age of podcasting? Podcasting? Golden age of podcasting? There's so much housecleaning to be done in the podcast industry before we can talk about golden age.

The economics of streaming with Will Page

14:34That is Will Page. I'm an author of the book Tarzan Economics, retitled Pivot as a paperback. Tarzan Economics, as in you have to let go of the old vine before you can grab the new one. And I'm the former chief economist of the Performing Rights Society, the equivalent of ESCAP or BMI in the United States, and then Spotify, a very famous Swedish music streaming service. And what did it mean to be chief economist at Spotify? So you would go from commercial negotiations work, willing buyer, willing seller, trying to establish a trade, between labels, publishers, telco deals, right the way across to artists work, artists who are furious at Spotify, who need to be convinced about the merits of the model in its early days to join the platform.

15:19Bands like Radiohead, who famously called us the last dying fart of a stinking corpse, winning them over to be a participant in the platform. Page was at Spotify when it went on that podcast buying binge. Not just production companies like The Ringer and Gimlet, but some podcast tech companies like Megaphone and Anchor. The economics is really looking at what would it bring to the platform. So if you invest in this bet, and this is a bet, some bets pay off, other bets don't, what's the retention like in the platform?

15:50Do you get people to join Spotify, which otherwise wouldn't? Yeah, fine with BBC Music. Oh, you got a podcast now? Now we're into that £12 a month subscription fee. I always think about streaming platforms as shopping malls, and what makes you park the car, get the kids out, get the trolley, and enter the mall. You know, that's what customer acquisition is about. So how did the bet pay off, and which components paid off well, which didn't, and overall, was it a good bet for Spotify? Well, I think it's not a taboo subject to say that the podcast bet has been watered down somewhat since those heady days.

16:22But I think the big ones were Anchor and Megaphone, investing in the tech stack. What Anchor especially, plus Megaphone, allows you to do is for Spotify to control the insertion of the advert, so that the platform has skin in the game. We also offer distribution models where imagery, GIFs, for example, can go with the podcast too. So modernizing the pipes of podcasting was definitely a bet worth investing in. Does podcasting look like a bubble to you in any way? It just needs to figure itself out. Just take a look at the casualties in the podcasting world, like the Wondery exit, like $300 million?

16:58Who got their fingers burned with that? Wondery was a podcast company started in 2016 by a former TV executive. It found huge success with narrative storytelling, especially true crime. Their idea was that a podcast could be just the first incarnation of a bigger media property, a movie or TV show especially, the same way that books traditionally have been. And it worked. Their hit podcast, Dirty John, produced with the L.A. Times, became a hit TV show. In 2020, just four years after it was founded, Wondery was sold to Amazon for about $300 million.

17:32Last year, Amazon restructured Wondery and laid off hundreds of employees. Spotify also had a big period of contraction after its buying spree. I remember working on the original Spotify podcast bet. And how an economist looks at things is different from the way that the lawyer or the commercial-minded person would look at things. One example was if you'd ask the industry, is YouTube a podcast platform? They would say, no, it's Spotify and it's Apple. They're the podcast platforms. Then you asked the consumer, what's your number one venue for a podcast?

18:05And they all said YouTube. That's a great example of where the economics kicks in, is the consumer and the creator are not speaking the same language. So what does the YouTube example tell you about what's knowable and not knowable, especially in a new industry? There's a lot of known unknowns in the world of podcasting. I'm fond of calling it the Wild West because there's so much noise out there that people are trying to make sense of. We reckon there's about 5 million podcasts out there. That's episodes, not shows. Spotify would estimate higher.

18:35Apple would estimate lower. That's interesting. What's really interesting is the scale of supply. So I calculated this for you this morning. Currently, there's a new episode going onto the unlimited digital shelf of Spotify and Apple every 0.9 seconds. What's the business worth? The highest estimate I've seen is $9 billion. There's lower ones as well. But one word or expression that I want to kind of land on your listeners to capture how I see podcasts is it's a sea of niches. I mean, there are big podcasts, but they're big niche podcasts.

19:08The confusion we're going to uncover in this conversation starts with that observation. It doesn't behave like other media industries have behaved in history. Profit is not necessarily the motive behind podcasting. Will, I have heard you say that you, and this is a quote, don't understand the economics of the podcast industry and that it, quote, doesn't make sense to you. So I was surprised to hear that because I thought if anyone should understand the economics of the podcast industry, it would be you. So please tell me what you don't understand. I'll come up with a term here called prune juice.

19:40Alan Sugar, an entrepreneur of some fame, bought Tottenham Hotspur Football Club. He said running a football club is like prune juice and that all the value goes around the sides of your body and out the other end. This is a graphic description for a podcast. Yes, it is. But if you think about the RSS feed where the adverts, the monetizable asset, is locked into the podcast, then where is the value for the intermediary in distributing it? Let me pause here for a second to focus on what Will Page just said.

20:12If the advertisements on a podcast, the monetizable asset, are locked into the podcast, which historically they have been, then what does the distributor get? By distributor, Page means Apple Podcasts or Spotify or the many other podcast platforms. And when he mentioned the RSS feed, that stands for Really Simple Syndication, which is an old but still reliable publishing format that gives each podcast a unique web address.

20:43And that's where the platforms can find the show and make new episodes available to listeners. RSS is a big reason why you can listen to a given podcast on nearly any platform. That's not the way it works for TV or music or pretty much anything. So if we're thinking about why podcasting is spread so far, so fast, it's partly because nobody really controlled it. That's one example of the Wild West of podcasting of just how can that be a functional supply chain?

21:14These were choices made by platforms. So Apple, for instance, rarely does anything for free. But Apple Podcasts was essentially a free service to the consumer, but also free to the producer. It's the greatest free lunch in media history. It could have been different, but it's stood the test of time. But let me just add to the warped nature of the supply chain. Let's say you're paying for one of the music streaming platforms, Amazon, YouTube, Spotify, Apple. And all you did last month was listen to podcasts. At no point did you stream copyrightable music.

21:46Do you know what happens to the 70% revenue share to record labels and publishers? I don't. Nothing. They still get their money. So there's no real incentive, monetary incentive for podcast consumption on the platform because it doesn't affect the gross margin structure of the streamer. When you step back and look at those two observations, A, nobody has to pay a penny to get to market, even though there's lots of costs involved. And B, even if you're paying $12 a month to Spotify and all you do is listen to podcasts, the record labels are none the wiser because they're still getting their 70% revenue share.

22:19You want to stand back at that and say, I don't understand the economics of podcasting. If the platforms are not making money off the podcasts, then what's their incentive to make their interfaces better, their discoverability mechanisms especially better? I would think there's zero incentive. Right, purely on the P&L, there'll be zero incentive because it just doesn't add up. Then we have to step back and look at the competitive landscape. What's the value of a moat? What's the value of acquiring a new customer?

22:49Spotify's move was to bring new customers in and keep them stickier. YouTube has its purpose as well. We're a huge player in the world of podcasts, perhaps the biggest of them all. YouTube is like the Death Star eating the world of media right now. But what is the value of a podcast? Zero. What is the value of a moat to protect this multi-billion dollar business? And that's what we have to kind of unpack here. I'm Stephen Dubner. This is Freakonomics Radio. We'll be right back. This podcast was born in 2010.

23:36It is still mostly a do-it-yourself project. I have a small production company that employs the producers and editors and engineers. And over the years, we've had a few different business partners for distribution and ad sales. In the beginning, we partnered with two public radio outfits, WNYC and American Public Media. Then we moved to Stitcher, a for-profit that was founded by a former WNYC executive. Stitcher ultimately got bought by SiriusXM, a publicly traded audio entertainment company.

24:08As the person who makes this show, I try to keep my distance from the business side, which I can do because of this person.

Representing top talent with Ben Davis

24:17My name is Ben Davis. I'm an agent at WME. Well, I was thinking, as much as I talk about the podcast business for over a decade now, I've never actually been on a podcast. Seriously? Yes, this is the first time. How can that be? No one's asked. I literally find that impossible to believe. Well, I'm not used to being on, Mike. I should say, you know, it's not our natural habitat as agents. Davis is a senior partner in the digital group at WME, and he represents some of the biggest names in podcasting.

24:50Brene Brown, Andrew Huberman, Malcolm Gladwell, Hot Ones, Jay Shetty. I was one of Davis's first podcast clients, and I'm pretty sure I'm his favorite. I don't have favorite clients. I think he's just saying that to protect everybody else's feelings. Anyway, I asked Ben to answer the question we got from our listener, Jerry. How does the podcast industry work, and where does the money come from? It's simply through ads, right? If this was a show on a cable network, they would sell the ads.

25:23They would pay their executives, pay for the real estate, and then negotiate with the host to pay them as little as possible so they can retain the upside.

25:32Yes, I'm familiar with that model. You're familiar. But in this case, you no longer need someone else's real estate to reach your audience. So you are the host, but also the owner and the network. And so those ads are sold by a third party, serious in your case. But then it goes to you and your company to invest in continuing to make this. And how are podcast ads priced? CPMs are the way that ads in general are priced. Cost per thousand impressions. And podcast CPMs tend to be around 17 to 22.

26:06Which is sort of back of the napkin for your average show. In general, those numbers have gone up. But the CPM for the biggest shows have gone up tremendously. If they're in a valuable category or tough to reach audience, they could go as high as $100 CPM. So for every thousand people, they pay $100. And it could go higher than that. How does CPM cost per thousand of ad sales in a median podcast compared to CPM in, let's say, terrestrial radio? It's much higher. Why is that? Well, one, the host read ads. So there's a relationship that you're lending your voice to read an ad, which of course exists on the radio side as well.

26:40But there's more of a direct relationship with the audience, such that the ads perform better, typically. I think that there's better tracking than many of those mediums. It's intimate. It's more pull than push, right? You're seeking out your favorite shows. And you're spending time with it week in, week out. I think that what the podcast industry gets wrong, it thinks too small about the world that it's living in. When you zoom out, this is really just self-distributed shows where we're going to see a lot of growth in the coming years. Our talent applying the podcast style model to new kinds of formats, formats that might have existed in television prior.

27:18When we see the big dollar deals, Dax Shepard, SmartList guys, Alex Cooper, Joe Rogan, Jay Shetty, if these numbers are in the, let's call it, $50 to $250 million, describe what goes into that and what comes out of it. Like, what's a platform getting for that, for a $100 million deal, let's say? What's the term of that deal? What's the frequency? How is the money earned back? Do those deals typically earn out, et cetera? So every deal's different, but typically these are multi-year deals.

27:49They could be three years, four years. And they're typically licensed, not sold, correct? Correct. These are rentals. So they're getting a fixed term where they have sales and distribution rights. Every single one includes ad sales rights. It might also include bonus content that's exclusive to one platform or another. It could include the obligation to do a certain amount of events over the course of the term, sponsorship rights to those events, social media posts. It's every surface that a podcast touches. Oftentimes, they include obligations on the talent to still deliver the goods.

28:21So these often include download minimums or impression minimums where the guarantee is contingent on delivering a certain amount of audience. Every situation is different. The leverage is different. The buyer is different. What share of podcast industry revenue is from ads versus subscriptions, donations, any other non-ad-supported model? I would guess around 80%, but there are some shows where it's the opposite, where they maybe are in a certain category or have a super loyal relationship with their audience.

28:52And they super serve them on Patreon with multiple episodes per week. And they make the majority of their money through subscription.

29:03In this way, the podcast business model has started to look more like the magazine or newspaper model or the newsletter model. A blend of advertiser revenues, subscription revenues, live events, merch, et cetera, et cetera. Here's Will Page again. This interview took place in June. We're now in a business where it's not how many downloads do you get, it's how much monetizable users do you have. Headline which just dropped today, not sure if you picked up on this, is that Patreon, one of the many sort of crowdfunding models out there, has announced they've paid $629 million to podcast us last year.

29:39And that's up by a third. Wow. What's that compared to the, well, the platforms pay zero, typically. Correct. How does that compare to the ad revenue that's going to podcasters? It's significant and it's going to the fewer as opposed to the many because, you know, there's a podcast every second going onto the platform. Not all of these can form membership groups. So, you know, I have the filtering, the bifurcation of the market. Joe Budden, he's announced that he's doing a million a month on Patreon. I mean, that's serious cabbage when you think about it. I think that companies have and continue to experiment.

30:11And Ben Davis again. What is the version of that where there's value that can be conferred on a platform where the audience become a serious subscriber or a Spotify user or whatever it might be? And there's been various experiments around how to do this in a way that makes sense. Initially, I would say it's sort of blunt force, but that will continue to evolve. And I think that there'll be ways that work. And it's really show by show. Do you ever, like me, find yourself thinking that the word podcast is just a horrible word and that it's too bad there wasn't something better?

30:43I totally agree with you. I think it's minimizing. And I think that it has this connotation which doesn't do justice to what people are doing. You know, I'm in a company where it's like I'm a movie agent. I'm a book agent. I'm a theater agent. I get to say I'm a podcast agent. I mean, frankly, I'm actually I say I'm a digital agent, which is I.T. to some.

31:06The way I choose to define it is it's a digitally distributed show. Nice. A digitally distributed show. This is Jim Bankoff.

Media consolidation with Jim Bankoff

31:16I'm the chair and CEO of Vox Media. Vox is a privately held media company that Bankoff helped build out of a network of regional sports blogs. With venture backing and then investments from big media firms, Vox turned into an acquisition engine. It now runs a vast collage of digital media properties and it has a substantial podcast division. We're just about 50 different shows that we run or 50 different groups of talent that we work with that produce shows.

31:48I think if you look at the audience size, it's roughly over half owned and operated shows. But if you look at the number of shows, it's predominantly partnered. I assume you don't give out revenues publicly? No, we don't. So can you tell me what share of revenues are from ad sales versus subscriptions and or other sources? In our case, as it relates to the podcast business, it's predominantly advertising. There are other sources and those other sources are growing faster than advertising. Those other sources include live events, include subscriptions, include license fees that we get paid by distributors.

32:25But first and foremost, it's advertising. If you look at the rest of our business, we have successful subscription businesses, notably around New York Magazine, which has over 400,000 paid subscribers and growing in a nice clip. And then we also have Vox and The Verge, which offer subscriptions and are just launched a little over a year ago and are doing really well. And then some of our podcasts also offer subscriptions or membership programs. All that news is good news for Vox, but it's not the big news.

32:55The big news is that earlier this year, they sold part of the company, including the podcast division, for a reported $300 million. Long story short, we split the company. Half of the company, roughly, will be owned by Lupa Systems, which is the investment arm of James and Catherine Murdoch. I will be the CEO of that company. That company will continue to be called Vox Media, and that company will be comprised of Vox. We sometimes call it Vox.com, which itself has great podcasts, including Today Explained and Gray Area and several other big podcasts.

33:33Also under Lupa will be New York Magazine. And very soon after selling off that portion of Vox Media, Bankoff sold the rest, including The Verge, SB Nation and Eater to Penske Media. That price wasn't disclosed. And it's worth noting that Jay Penske's company had already invested in Vox $100 million back in 2023. We've been working closely with them as a partner for many years, and they got to know the business very well. Jay has a lot of different businesses that he manages, but one of them is what I'll call this publication business.

34:08They have an outstanding portfolio as well. They have Variety, Rolling Stone, Hollywood Reporter, Rob Report. Their list goes on. And so the idea is to take their publications, create a new company that encompasses theirs and ours. So Vox Media back in 2015 had a valuation for whatever that's worth of about a billion dollars, I understand. And now you just sold the podcast and magazine parts for, let's call it $300 million. That seems like a, I don't even want to say a devaluation, it's just a different animal.

34:41If you especially compare the price you got for Vox Podcast to what Spotify paid for the three production companies that they bought at the time, Gimlet, Parcast, and The Ringer, what does that say about the valuation of podcast networks or producers themselves? Does it feel like that was bubble and now we're normal? Does it feel like it was a crash from which there won't be a recovery? Does it feel like this is another early day in what you think will be a longer term rise in valuation?

35:13How do you see it? I'm not going to be the first to say that valuations ebb and flow just like the stock market does. I mean, listen, we're very privileged and fortunate to build something that commands so much value in so many people's eyes because of all the hard and smart work that goes into it. But what we really want to do is continue to have an economic model that enables us to grow and create outstanding work. These transactions are focused on the long term.

35:45I've always been a little bit surprised that Apple, which likes to make money, Apple is an amazing company that's really, really good at making money by making stuff that people really want. But from the outset, Steve Jobs just kind of decided that it would be cool to have a podcast app on the iPhone and instantly created this global instant free distribution system for someone like me, which is a remarkable gift. I mean, it reminds me of in the early days saying I'm going to build railroads everywhere around the world and anybody who wants to can put their trains on them and take them anywhere like that doesn't happen except in the digital world.

36:24But Apple decided not to really try to get a piece of the podcast action. What do you think might have been different about the industry if they had? Yeah, that's an important question. I think it would have stifled growth and adoption or it would have at minimum left the door open for other entities to come in. Podcasting grew up on an open distribution system, much like the web grew up in an open distribution system.

36:54There have been movements to close the web ecosystem. Well, there were movements to close the podcast. And there are movements to close the podcasting one over time. Why do you think those efforts failed, Jim? Because these were the big players, Spotify, Amazon, and so on. After acquiring a podcast studio, they would make some of their shows exclusive to their platforms. You couldn't find them on Apple or elsewhere. Ultimately, they reversed that walled garden approach. But why do you think it didn't work? One thing I've learned over time is the platforms are going to operate in their best interest.

37:29And it's in their best interest to have a wide variety of content out there, a wide variety of creators contributing to their platform. And an open ecosystem encourages that more so than a closed ecosystem. I keep reading that within the podcast economy, there's something like a billion dollars of advertiser demand sitting on the sidelines because the industry can't agree on how to measure audience exactly. Number of listens or number of listeners or the share of an episode that is consumed and so on.

38:05Can you just talk about measurement itself as an issue? I think the industry and frankly, the platforms need to do a better job. YouTube might have one way of measuring Spotify, Apple, RSS. There's a lot of different formats. And so it's often left to us to stitch them together and be direct and honest with our advertisers about what's going on. If anything, I think we're all being undervalued right now. And I look forward to helping fix that problem. When you realize there's no real audience measurement figures still to this day, and we've had podcasts for over 20 years, it's the Wild West.

38:43That again is Will Page, former chief economist at Spotify. Are we going to stick to this RSS feed, which has served podcasting since 2004? Are we going to adopt something new? Are we going to stick to these flaky audience figure measurements? Are we going to adopt something new? Are we going to stick to multiple subscription models, Patreon, Kickstarter, or are we going to have a common currency where I can subscribe to three or four podcasts in a simple click? Can I say, if we just jump across the media verticals here, it's not dissimilar from the book industry. You know, why can't I go to Amazon and in one press, one transaction, own the hardback for my shelf to impress my friends, the Kindle so I can read it on a beach, and the audio book so I can hear it on my commute.

39:22We've had Amazon books since 1997, and I still can't do that. That's not consumer-centric at all. I mean, is that the first step, and then a kind of more organized industry can coalesce around it, you're saying? Yeah. I think we need to let go of the old vine of the amateur podcast and reach out to the new vine of the professional podcast. Separate out the filler from the killer. And that's going to involve some hard choices for an industry that's been brought up on the RSS feed, brought up on hobbyism. You want skin in the game. If you want the platformers to take this format seriously, I think the industry itself has to take itself more seriously.

39:57And I think you'll see more walled gardens, more exclusive arrangements, more investment in video, more competition with television. This summer, Oxford Road, one of the biggest podcast ad agencies, convened the AMP Accords, the Alliance for Measurement in Podcasting. Their main claim is that podcast advertising measurement is broken. Our agency spends more than half a billion dollars a year inside podcasting, Oxford Road CEO Dan Granger said. And I still cannot tell a chief marketing officer with confidence what counts as an impression.

40:32Television solved that in 1950. Digital solved it in 2005. Podcasting is the only major medium left where the platforms ask advertisers to take their word for it. And advertisers still do.

40:48So the podcast industry is trying to mature, let's call it. But how long will that take? Because something else is already happening. Two large language models walk into a bar, okay?

41:03What does AI do to podcasting? That's coming up after the break. This is Freakonomics Radio and my name is Stephen Dubner. I appreciate your listening.

Automating content with AI generation

41:25Over the past year, the podcast ecosystem has been flooded with AI-generated content. Here again is Will Page, former chief economist at Spotify. It was already easy to get to market as humans. It's so much easier to get to market with so much more volume when you're just AI slop, when there's no human in the mix. So let's imagine an AI-automated show produced in real time. It could be a tragedy or a big event in the news. It happens so you can knock out that podcast before any human can get to market.

41:58Let's say it gets just 200 plays, okay? And you've got a $20 CPM on those 200 plays. That means you've earned, mathematics here, $4. And let's just slap a marginal cost down there of $3 in production costs. So that means you've made a dollar profit. And it's the ability to flood the market because plays mean money. It's got nothing to do with retention. So is this a substitute or a complement to human-made podcasts then? So this is where it gets quite spicy. Dilution theory is real.

42:29A song lasts for four minutes of my time, a podcast lasts for 40. AI-slot podcast, if it lasts for 40, is 40 minutes that I can't give a human-created podcast as well. So it has real effects in terms of dilution because it's competing for so much of the clock. The big discussion point in the world of podcasting is, of course, in section point AI. My name is Janine Wright. I am the co-founder and CEO of a company called Inception Point AI. We make AI-generated personalities and AI-generated content that those personalities release into the world.

43:02And how much content is Inception Point making? Here's what Wright told me when we spoke in May. We have about 15,000 active shows, more than 300,000 active episodes. But it seems like all those podcasts aren't really the point. We think of ourselves as a tech company, not a media company or as a podcast network. We have built the underlying platform that allows us to work with other third-party companies to allow them to make AI-generated content and AI-generated personalities. Janine Wright has been in the podcast industry for a while.

43:35She was co-founder and COO of Simplecast, a hosting and distribution platform that was acquired in 2020 by SiriusXM. And she was COO of the podcast network Wondery that we mentioned earlier. But she sees Inception Point AI differently. Right now, the overwhelming majority of the content that we're making is in audio, but we are also in video. We're on YouTube and the social media platforms. And we're also in beta on real-time personalities. So being able to talk on the phone, text, live chat with our personalities.

44:09Okay. Tell me where your funding comes from. Yeah, so we raised $5.2 million in our seed series. We raised quite quickly in an oversubscribed round. And that was our seed funding in order to really give us an opportunity to be able to build what we believe is the platform for the future. The way that we think most content will be made in the future is AI-generated content and AI-generated personalities. You mentioned 15,000 active shows, 300,000 episodes.

44:40Give me a sense of listenership. Maybe take your five biggest shows. Tell me how many downloads or streams or listens or however you want to measure that. Each one gets in, again, whatever window of the metric you want to do. One day, 30 days, 90 days, etc. That's, I think, an interesting question because it's kind of what we believe is fundamentally broken about the business model for many traditional media companies. Is you become very dependent on a very small number of hit shows. It works much more like a VC portfolio where you're hoping that you have a few big ones that break out and you invest a ton of production and marketing resources in trying to get those to break through.

45:20But because we have really radically reduced the production costs, we think about it on a portfolio basis. So we're now reaching more than about a million downloads a month across our portfolio. But the overwhelming majority of our shows are small in comparison to what you would traditionally think of as a hit or a success in the podcasting business. So no offense at all, but a million downloads a month across a portfolio of 15,000 shows is, you know, terribly low compared to established podcasts.

45:51So I understand the portfolio argument, but when VCs invest in a portfolio of assets, they're investing quite a bit of money or at least, you know, enough money for the startup founders to do what they need to do, which is, you know, real estate, hiring, the tech stack, etc., etc. Your costs, though, are astonishingly low. It seems close to zero. I've read that you make an episode for about a dollar. Is that right? Yeah, so on an aggregated basis, it comes out to about a dollar an episode.

46:22I get your point with respect to the business model and the number of listeners that we're reaching. I mean, first and foremost, the business model is not designed around monetization of our own content. The vision is that we're building a tech company that will make the overwhelming majority of our money from our relationships with enterprises and enabling them to be able to use our platform to do the same. Is there advertising on your shows? Yeah, I mean, you've caught us in a little bit of a weird time because we're migrating platforms and so we're getting the new systems set up.

46:52But yes, previous to this, we have been supported by programmatic advertising. For people who don't know what programmatic advertising is, just describe that quickly, please. So that's advertisers who are not looking to advertise in any particular show or with any particular host. You know, so it's not somebody who went to sign up to advertise on Freakonomics or to have Stephen Dubner read an ad spot or endorse a particular product. Just to be clear, I don't do endorsements, but thank you. Okay, sorry. No, that's okay.

47:22I'm a little Boy Scout-y about that. Yeah, so these are advertisers who have said, we just want to reach particular kinds of audience. So we want to reach audiences that like audiences that live in California or audiences that might like outdoorsy kind of content, right? So they're buying audiences in big buckets of content and they really don't have a particular reference as to the shows that make up that bucket. We think that content in the future will be more niche, more micro-community, more personalized.

47:53So we do pollen reports and ski reports and local music scenes, all sorts of things about weather or news or local community updates. And then when we find markets that work, then we roll them out into a hundred different markets. Yeah, where's the information coming from? How is it turned into content? And then how is that content meant to be found by a would-be consumer? So we are not a model maker. We're an AI orchestration layer. So we are pulling from a number of different LLMs and we're pulling from a number of different data sources, depending on whatever the topic is that we're creating content around.

48:29If we're going to do, for example, a local crime report, there is local data sources that are available that are very influential in that content. So we'll pull from a direct data source. A police report, for instance, in that case? Police report, court filings, even sometimes publicly available police blotters, all sorts of things like that. Let me go back to your pollen example. What about fact-checking and any other version of something like fact-checking to assure that you're actually disseminating the information from the right place, the right date, with the right intensity, with the right nuance, etc.?

49:05How do you think about that? Yeah, so that's a lot of what I think is the secret sauce of what we have built. Nowadays, you're seeing it's pretty easy for somebody to just go to, you know, Notebook LM or even today you can go to your Alexa and say, you know, make me a podcast. And the next 10 minutes later, we'll make you a podcast. We have built into our system various QA processes throughout. QA meaning quality assurance? Quality assurance that allows to do a bunch of gate checking. How much of that is done by humans versus other AIs?

49:38None of it is done by humans today. Half of my team is creative. So, yes, we have creators. The creators spend the overwhelming majority of their time at the very tip of the sphere, the front end of the process. They're designing the personalities. They're designing the formats and the recipes. And they're figuring out what the story arcs are going to be. And then they are reviewing and making sure that these processes at the beginning are producing the quality of the content that we want. And then we have our tech systems that come in and automate everything about the process that it possibly can.

50:12With 15,000 shows, I would think there's at least, you know, a few hundred humans there involved in this whole creating, checking, generating, disseminating, monetizing, et cetera, et cetera process. How big is your staff and payroll? No. We just hired our 11th employee.

50:29So there's about five tech folks, five creators and me. We also have eight agentic employees we've created that work for us full time in service of the work that we're doing. Why are they called employees if they're agentic? They have names. We assign them roles. They have specific tasks. Are they paid? They are not paid. No. They don't have benefits. We pay somebody for them. They don't have benefits. But being able to leverage these agents to help work for us helps make us more efficient and makes it so that, you know, we just rolled out our first 401k.

51:01Congratulations. Thank you. Actually, I referenced your nudge theory episode when I rolled it out. Got it. Plainly, you're building a company around the idea that AI-generated content of a fairly high level is legitimate and will have an audience and will be valuable and most of all make a lot of money. Tell me what persuades you that's actually true. Yeah, I mean, I think it's hard for human brains to even comprehend the amount of data that is going into these underlying models that are feeding the information that we're then using to generate content.

51:34And being able to check one model's outputs against other model's outputs really makes those outputs then in and of themselves more reliable. I think it's also, you know, the fact that this is the most invested in technology of all time. You literally have millions of people hours, hundreds of millions of dollars going into improving this technology every single month. And our processes are getting better at leveraging the technology over time. So you've got one show called Jason and Travis Kelsey Football Brothers.

52:07There's an image, which I assume is AI-generated image. Jason looks pretty much like Jason. Travis looks not that much like Travis. How would you describe that show? What's the need, what's the audience for an AI-generated version of a Travis and Jason Kelsey show? It's people who are wanting to have regular information about those people. And we help provide that information by aggregating information that is news and updates that have happened about these folks and then putting it together with a personality in a storytelling type way that people like to consume.

52:42So I'm curious, has there been legal action either threatened or taken by the Kelsey's or has there been legal action taken or threatened by anyone else who is the underlying subject of podcasts that you created with AI? I mean, not really. Every once in a while, we'll get somebody who doesn't like the news coverage about them. We do a lot of stock trackers, for example. And so when a company's stock is declining, sometimes we'll get a brand that will reach out and say, oh, we don't love this stock tracker that you're doing about this.

53:13Could you not use our brand name? What do you do in that case? Most of the time we change it or we take it down. When we're operating at this kind of scale, even if we believe that we're squarely within our rights because this is a public figure or the newsworthiness protections, it's not worth it for us on a single show basis to contend. What happens to your business model if the underlying LLM providers are successfully sued for how they've ingested data from copyrighted sources?

53:43We rely on the data that we're receiving from our providers. So if they are sued or they engage in or enter into, for example, license agreements with their parties to provide content, then their costs in order to deliver the content to us go up and our costs go up as a result. It seems to me that humans love technology, but they love humans even more. Do you remember when Deep Blue, the computer made by IBM, beat Garry Kasparov in chess? And it was this amazing technological breakthrough and people who understood chess were knocked out by it.

54:18And the predictions were that human chess was finished. You know, why bother when the computer will always win? And yet it didn't happen at all. People still overwhelmingly prefer watching humans play humans as opposed to watching a human play a computer. Certainly two computers play each other. So with an AI-generated podcast where the content is coming from a pool of publicly available knowledge, essentially, and where the voice and the intentions are different than the human elements that we're accustomed to, what makes you think that people want what you're making?

54:57I think that we're in an interesting moment in time where people want there to be two different categories. They want there to be AI-generated content and not AI-generated content. And that line is becoming very blurry because more and more people are starting to use AI in all sorts of different aspects of their content creation processes. And then you start getting into this sticky situation of trying to find out what is it that creators and what is it that consumers actually care about being AI-generated, right?

55:33Is it really that they have a problem with a synthetic voice or is it they have a problem that the script is generated in an LLM or that the production process has become more automated and there's not, you know, a real-life person in there or that there's no real-life fact-checker or real-life researcher? And when you poll a crowd, I think you would get very different answers across the whole spectrum.

55:59Yes, I would agree with Janine Wright on this point. When it comes to content that is primarily generated by AI, yes, you will get very different answers across the spectrum. For instance, at the opposite end of the spectrum from Janine Wright is the podcast producer, Jenna Weiss-Berman, who we met at the start of this episode. I asked her if she is familiar with Wright's company, Inception Point AI.

56:28Yeah, of course. I think they're the devil. And I'm happy to say that. Well, you just did. So I hope you're happy. Yeah, because what are we doing here? Like, what are we doing? Art is art. Like, this medium is creative. Why are we trying to bring in artificial intelligence? Stephen, artificial intelligence? Aren't you more interested in human intelligence? In actual intelligence? Also, like, the friction that is needed to make good and creative and interesting stuff?

56:58I don't understand why I'm supposed to be excited about something that has the potential to kill a ton of jobs. I love working with humans. I love hearing their ideas. Like Amy Poehler. Like, AI can't copy Amy. What are we doing?

57:17A few weeks after I spoke with Janine Wright, the CEO of Inception Point AI, the firm rebranded itself as Liquid Studios. Here's what Wright told us by email. The biggest lesson we learned was that betting on one model or even a handful is a death sentence. She said that Liquid Studios will still be using AI tech, but it will be directed toward existing content. We don't start from a blank page anymore, she wrote. We start from something a human already cared enough to make.

57:48Jenna Weiss-Berman, meanwhile, has just launched a new, very human-centered podcast with Ina Garten, the cookbook author and longtime host of Barefoot Contessa. It's called Happy Hour. She'll be mixing drinks and explaining how they're made, chatting with fun, interesting guests. And she's just so sweet and lovely. And was this your idea or her idea? We started talking about it two years ago. I always say, you always got to take two years. Two years for me seems like nothing.

58:21I've been making Freakonomics Radio now for 16 years. Would happily take another 16. This project has allowed me to learn so much, to speak with so many interesting people, to solve so many puzzles. So, I think podcasting is the best invention in the history of the world. I do hate the name, though. I went back to my friend and agent, Ben Davis, for one last try.

58:49Do you have any suggestions for better names? Have you ever heard anybody refer to their podcast as something different that you think was promising? Um, shows. These are just shows, right? Wait a minute. Have I been in show business this whole time and I didn't know it? You're in show business, baby. Yes, come on. My thanks to Ben Davis, Jenna Weiss-Burman, Janine Wright, Jim Bankoff, and Will Page for today's conversations.

59:19Also, thanks to our listener, Jerry, for the idea. And since I now know that we are in show business, maybe it's time to take you behind the scenes. We're going to try a new occasional episode called Ask Us Anything. You can send an email with any kind of question or observation or complaint about this show and we will answer it. Our email is radio at Freakonomics.com. In the subject line, write Ask Us Anything. You can ask about how an episode actually gets made, why we asked this person that question, or why we didn't,

59:57why we haven't followed up on something you've been waiting eight years to hear about. So, again, ask us anything, questions, complaints, old grudges, they're all welcome. Write to radio at Freakonomics.com. Meanwhile, coming up next time on the show, David Rubenstein seems to be what you might call a virtuous billionaire. I got everything I need. I'm just really trying to give back to society. I got very lucky in life. My parents were not college educated. They weren't even high school educated. His list of philanthropic involvement is exhausting, although he is no longer chairman of the Kennedy Center.

1:00:34I'm the only person ever to be fired by a president and succeeded by him at the same time. And now Rubenstein owns a baseball team. That's very frustrating when you lose because you feel like you're responsible for it, but you can't really make many changes on it. That's next time on the show. Until then, take care of yourself. And if you can, someone else, too. Freakonomics Radio is produced by Renbud Radio. You can find our entire archive on any podcast app. It's also at Freakonomics.com, where we publish transcripts and show notes.

1:01:06This episode was produced by Mandy Gorenstein, was edited by Ellen Frankman, and mixed by Jake Loomis with help from Jeremy Johnston. The Freakonomics Radio network staff also includes Dalvin Abouaji, Claire Keenan-Kurgan, Eleanor Osborne, Elsa Hernandez, Pete Madden, and Teo Jacobs. Our theme song is Mr. Fortune by the Hitchhikers, and our composer is Luis Guerra. As always, thanks for listening.

1:01:34Here's my favorite podcast joke. What's the best way to keep a secret, Stephen? Put it at the end of a podcast? Correct. Correct.

1:01:43The Freakonomics Radio Network. The hidden side of everything.