Everyone Has a Podcast Now: A Roast of the Three-Microphone Era
There is a new unit of measurement for modern friendship: how long someone can know you before mentioning their podcast. The answer, increasingly, is eleven minutes. The three-microphone era has arrived — three friends, three matching mics, one ring light each, and a ninety-minute conversation that will be clipped, captioned, and released into a feed already groaning under the weight of identical furniture. Everyone has a podcast now. The roast practically writes itself, which is convenient, because so does the podcast. But a fair roast checks the numbers first, and the numbers are funnier than the jokes: the flood is real, the audience for almost all of it is not, and most of the flood has already quietly drained away.
The census of the graveyard
Start with the headline figure the launch posts never quote in full. Podcast statistics trackers citing the Podcast Index count roughly 4.73 million podcasts indexed worldwide. Impressive — until the same trackers subtract the inactive shows and land at approximately 450,000 to 500,000 podcasts actually producing content. More than four million indexed shows are, in the industry's own cheerful term, podfaded: launched, loved briefly, and abandoned. The internet does not have a podcast problem so much as a podcast archaeology problem. Nine out of ten artifacts in the dig site are already fossils, each with a trailer episode in which two hosts promise, with total sincerity, that new episodes drop every Tuesday.
The fresh-launch data is bleaker and better. Podchaser analysed all 333,723 podcasts that launched in 2025 and checked which had published anything in the previous ninety days. Roughly three in four had gone quiet. A full 26% never published a second episode. The median show that stopped lasted three episodes and eight days. Eight days. There are houseplants with longer public careers. The traditional wisdom said podcasts die around episode seven; the data says the cliff is episode one, and most of the era's output is not a catalogue — it is a very large pile of first dates that never called back.
Three eras of the medium, one of them wearing a costume
How did a medium built for the ears end up as a furniture arrangement? A short, unfair history:
| Era | The setup | The tell | The failure mode |
|---|---|---|---|
| The bedroom era | One host, one cheap microphone, a topic they could not stop thinking about | Audio only; the show existed because the subject would not leave the host alone | Burnout, eventually — but the episodes were about something |
| The network era | Studios, producers, seasons, budgets | Edited, structured, and occasionally fact-checked | Expensive; cancelled like television, because it was television with worse lighting |
| The three-microphone era | Three hosts, three mics, three cameras, a couch, a neon sign with the show's name | Video first; the conversation is the raw material for clips | Podfade by episode three, survived by the neon sign, listed on a resale site |
The third era's defining move is the costume: the set is dressed as a talk show, because a talk show is what the format is imitating. One newsletter critique puts the secret plainly — many podcasts are just poorly executed talk shows by another name, and the giveaway is that the hosts often do not listen to their guests, missing the subtext, steamrolling the answers, waiting for a gap to deliver the take they arrived with. A great talk-show host responds to the energy in the room. The three-microphone host responds to the clip potential. The mic count went up. The listening did not.
The argument the internet is having with itself
Search the question and the results immediately start fighting. In one corner, the scolds: a representative student-paper opinion argues that podcasting is popular because people love to talk and shows are simple to start — friends have launched them by recording FaceTime calls — and that the resulting oversaturation means quantity now outweighs quality, with many shows circling the same ideas. In the other corner, the coaches: a podcasting coach replies that nobody asks whether there are too many books or songs, and that most existing podcasts are dead anyway — the coach's own figures claim 90% of podcasts never get past episode three, and most of the rest quit by episode twenty — so the field is emptier than it looks and a new show only has to serve its own specific audience.
Both corners are half right, which is why the argument never ends. The scolds are correct about the texture: the feed genuinely is flooded with interchangeable conversation. The coaches are correct about the arithmetic: the active field is a tenth of the headline count. What neither side says out loud is the reconciliation — the flood and the graveyard are the same shows. The era produces a torrent of launches and a trickle of survivors, and the torrent is what listeners wade through while the trickle is what they actually hear.
Should you start one? The survival-data test
Podchaser is careful to call its survival findings associations, not cause and effect — but associations from a third of a million launches are the best map available, and they make a sharper decision framework than either the scolding or the pep talk:
| Question | Green light | Neon-sign light |
|---|---|---|
| Can the format run without you inventing a topic weekly? | Guests carry the show — among shows reaching 26–52 episodes, 67% of guest-booking shows were still active versus 50% without guests | The concept is "we talk about stuff" — the exact profile of the eight-day median lifespan |
| What cadence survives contact with a real life? | Biweekly had the best survival rate of any cadence at 44% | Daily came last at 28% — ambition is a schedule's worst enemy |
| What does episode one prove? | Nothing yet — 26% of launches never reach episode two, so the plan must make episode two boringly easy | The launch is the plan; the trailer promises Tuesdays forever |
| Why video, honestly? | A specific clip strategy for a specific audience | Because the couch-and-neon set is what podcasts look like now — a costume in search of a show |
Pass the green column and the roast withdraws its objection: a guest-driven, biweekly show on a subject with a waiting audience is not part of the flood — it is part of the surviving tenth. Fail it, and the kindest available act is to have the ninety-minute conversation anyway, enjoy it enormously, and release it to exactly the group chat it was always for. The three-microphone era will produce its millions regardless. The only say anyone gets is whether their show is census or survivor — and the data, unusually for a roast, is happy to help decide.
FAQ
How many podcasts are there, really? About 4.73 million indexed worldwide by Podcast Index counts, but only roughly 450,000–500,000 actively producing — the rest have podfaded.
What is podfade? The industry's term for a show that quietly stops publishing. In the 2025 launch cohort, roughly three in four shows had gone quiet within the check window, and the median casualty lasted three episodes and eight days.
So should nobody start a podcast? The data argues the opposite of both extremes: launches die fast, but serious shows with guests and a sustainable cadence survive at far higher rates. Start like a survivor, not like a launch.