The Podcast Nobody Listened To, For Fourteen Months

Tomás recorded his first podcast episode on a Saturday in February, in a closet in a flat in the 105 postcode of Reykjavík, with a blanket nailed to the back of the door for sound dampening. The blanket had a reindeer on it. He'd bought it at a tourist shop two years earlier as a joke gift for his sister and never actually given it to her. The microphone was a used Shure he'd found on a Facebook group for fifteen thousand krónur. The other microphone, the one for the guest, was a worse Shure he'd bought new because the used one had a dent.

The guest on episode one was a friend of his named Pétur who ran a small invoicing SaaS for Icelandic carpenters. They recorded for an hour and twelve minutes. Tomás edited it for six hours over the following weekend, listening back to every section three or four times, removing every "um" he could find. He gave the episode a piece of intro music he'd licensed for twelve dollars on a stock site — a sort of upbeat synth thing that sounded like a podcast about productivity hacks, which the show was not.

He published it on a Wednesday morning at 7 a.m. Reykjavík time, because he'd read somewhere that 7 a.m. was the optimal podcast release time, although he could never find the source again. He sat at his kitchen table with a coffee and watched the download counter, which is something you can only do once in your life because after that you know better. By the end of the day, the episode had four downloads. Two were him, checking on it from his phone and from his laptop. One was Pétur. The fourth was his mother, who he was pretty sure had not actually listened.

That was February. The plan was two episodes a month, interviews with indie founders. He kept that schedule, more or less, for fourteen months. Twenty-seven episodes. The average download number, by month eight, had settled at exactly eleven. He knew it was eleven because he checked. He also knew that of those eleven, four were him listening back at different stages of editing, on different devices, to check audio levels. The actual number of human strangers listening to any given episode of his podcast was, on a generous estimate, seven.

He kept going anyway. Not for the seven. For the conversations. There was something about the act of sitting across from another founder for an hour with the recorder on that made them tell him things they would not have told him at a meetup or in a Twitter DM. He was, accidentally, getting a free education in how a couple dozen indie businesses actually worked. He loved it. The download number was a tax he paid to do the thing he loved.

Episode 28

Episode 28 was recorded on a Thursday night in early April, which is already a break from the format because Tomás recorded every other episode on a Saturday afternoon. The guest was a woman named Helga who ran a one-person scheduling tool for Nordic hairdressers. She lived four blocks away. They were friends — actual friends, the kind who borrow each other's drills, not the kind who follow each other on LinkedIn. She came over after work. They had a beer first, which they had never done before recording, and then they sat down in the closet with the reindeer blanket and Tomás hit record.

About four minutes in, he realized he hadn't pulled up his question list. He always had a question list. He'd spent two hours that afternoon writing it, the way he always did, with twelve questions in a Google Doc, ordered from "warm-up" to "meaty" to "reflective." The doc was open on his laptop, but the laptop was in the other room because Helga had said the keyboard sounds were distracting on her last interview. He thought about getting up to grab it. Then he didn't.

They talked for two hours and ten minutes. About forty minutes in, Helga started telling him about a SaaS she'd built before the scheduling tool, a thing for veterinarians, and how it had failed. She didn't tell him in the way founders tell failure stories in interviews, with the polished arc and the lesson at the end. She told him in the way you tell a friend, with the bit about how she'd cried in a parking lot in Akureyri after a meeting with a vet who said the product was "fine, I guess." That tangent went on for fifteen minutes. He didn't redirect. He just asked one quiet follow-up question, and then another, and then she kept going.

When they stopped recording, he knew it was different. He could feel it in his chest, the way you can feel a good conversation in your chest after a dinner that went on too long. He didn't know yet that the episode would do anything. He just knew it was a better thing than the twenty-seven episodes before it.

He sat with the audio for two days before editing it. When he did, he made a series of small decisions that he had not planned. He cut the intro music. The episode just started, with Helga mid-sentence saying "…and the worst part is I actually liked the vets, you know?" He didn't write show notes — he just put one sentence in the description: "Helga and I talk about her dead SaaS for veterinarians, among other things." He didn't trim the long pauses. There's a pause in that episode, around the 53-minute mark, that lasts almost nine seconds. He left it.

He published it on a Tuesday, not a Wednesday, because he'd finished editing on a Monday night and didn't feel like waiting.

Episode 28 hit nine thousand downloads in three weeks. Not nine. Not nine hundred. Nine thousand.

Tomás did not understand this for several days. He kept refreshing the dashboard the way you refresh a weather forecast when the number is one you've never seen before. The previous twenty-seven episodes did not move. They sat at their eleven downloads. Some of them went up by one or two as people who discovered episode 28 went back to sample the archive, but the archive did not retroactively become a hit. Only episode 28 was a hit.

He had shipped no new marketing. He had not tweeted differently. He had not gotten a feature in a newsletter. There was no post on Hacker News, no mention from a famous founder, no algorithm-bait reel. He looked, several times, for an external cause. There wasn't one. Or, more precisely, the cause was the show itself. People were sharing it because the show was different. The episode was the marketing.

He spent the next month trying to understand what had actually changed. Not so he could repeat the trick — he knew, instinctively, that trying to repeat a hit was a different and worse project than trying to make a good show. But because he wanted to know what he'd been doing wrong for twenty-seven episodes.

The format trap

Here's the thing he figured out, and I'm going to make it bigger than his story because his story is one of a few hundred I've seen that rhyme.

Most indie podcasts copy a format from a podcast they already love. The founder who listens to NPR makes a podcast that sounds like an NPR show, with scripted intros and tight edits and a host voice that's a little too polished. The founder who listens to Lex Fridman makes a podcast that's three hours long with a slow zoom on a black background and the same opening jingle. The founder who listens to Lenny's Podcast makes a tightly-produced interview show with a question list and clear segments and a guest bio at the top. None of these formats are bad. They are, in fact, very good, for the people making them, at the scale they make them at, with the production budgets they have.

None of those formats fit a Saturday afternoon recording in a closet in Reykjavík.

Tomás's first twenty-seven episodes were trying to be a small version of a big show. He had the intro music because NPR has intro music. He had the question list because the interview podcasts he admired had question lists. He had the clean edits and the show notes and the "before we start, if you're enjoying this, please subscribe" line because every podcast he listened to had those things. He was making, in effect, a low-budget knockoff of a format that was designed for a different kind of show, with a different host, and a different relationship to its audience.

Episode 28 wasn't a small version of a big show. It was the actual size of his actual show — two people in a closet, talking. The format finally fit.

The format has to fit the speaker

I want to put this one as plainly as I can. The format is not a genre. The format is a piece of clothing, and it has to fit the body of the person inside it. A two-mic Saturday podcast hosted by an indie founder who has never done radio is not the same body as a former public radio producer with a studio in Brooklyn. Putting the public radio format on the indie founder body looks exactly as awkward as you'd expect. We just don't see it because we've been trained to think the format is the show.

It isn't. The host is the show. The format either makes the host more like themselves, or it makes them less like themselves. There is no neutral format. If you are choosing one, you are choosing to amplify some part of your voice or to suppress it. Most indie podcasters, without realizing it, pick a format that suppresses.

Two friends with a recorder is harder than it looks

The trap on the other side is to read what I just wrote and decide that the answer is "just press record and have a conversation." If only. The reason episode 28 worked was not that Tomás stopped doing the format. It was that the conversation underneath the format was real. He and Helga were actually friends. They had actual stakes in each other's businesses. Helga had cried in an actual parking lot in Akureyri.

If you put a recorder between two people who don't really know each other and tell them to "just have a conversation," you do not get episode 28. You get an awkward podcast where two strangers perform friendliness at each other for an hour, and the audience can hear the performance, because the audience can always hear the performance. You also need the willingness, as a host, to leave the long pause in the audio. Most indie podcasters cut their pauses because the pause feels like a mistake. The pause is the show. The pause is the thing that says, to the listener, "this is a real moment that I am not going to sand down for you."

The math of eleven downloads

I want to spend a minute on the numbers, because I think small podcasters carry a kind of quiet shame about download counts that is almost entirely the result of bad math.

Eleven downloads is not eleven listeners. Tomás knew this because he'd done it to himself — four of his eleven downloads were him, on different devices, at different stages of editing. If you assume his other listeners behaved roughly the same way (some of them re-downloaded to listen on a walk, some of them had their podcast app on three devices, some of them re-downloaded after a phone wipe), eleven downloads is probably four to six actual humans.

Four to six humans is not a failed audience. Four to six humans is a dinner party. Most podcasters would, if asked, kill for a dinner party of four to six engaged listeners that they could call by name. They would, in fact, pay for that. What they instead do is treat that exact dinner party as a humiliation, because a dashboard told them the number was eleven and a different dashboard told them some other podcast had two million.

The dashboard is lying to you, in both directions. The eleven is smaller than it looks, in raw human terms — it's maybe five — and bigger than it looks, in relationship terms, because those five people are the kind of people who will listen to a closet podcast for an hour. They are pre-qualified weirdos in the best possible sense. You should know their names.

The number you should obsess over is not downloads. It is whether anyone messages you about an episode. One unsolicited message — a real one, with a specific sentence about a specific moment in the show — is worth more than four hundred downloads from people who autoplayed the episode while doing the dishes and never thought about it again. The message means the show landed. The download means the file moved.

Here is a rough sorting of the kinds of listener signals a small podcast actually generates, in ascending order of how much they should matter to you.

• The follow. Someone hits subscribe in their app. You will probably never know they did this unless your hosting dashboard tells you, and even then the number is delayed and noisy. Worth roughly nothing on its own.
• The download. The file moved from your host to a device. It does not mean the episode was played. Many podcast apps download episodes automatically the moment a new one drops. Worth slightly more than nothing, in aggregate, useless one at a time.
• The completion. A listener got to the end of the episode. Some hosts report this; most don't, accurately. When you can see it, it's a much better number than downloads, because finishing a ninety-minute conversation requires actual commitment.
• The reply tweet or the screenshot share. Someone quoted a specific sentence from your episode in public. This is the first signal that means a real human did real listening. It also costs them social capital to do, which is why it counts.
• The unsolicited message with a specific timestamp. "I loved the bit at 47 minutes where she talked about the parking lot." This is the highest-resolution signal a small podcast can produce. One of these per episode means the show is working, regardless of what the dashboard says.

Six format breaks worth trying

After episode 28, Tomás stopped doing a lot of things he had been doing for twenty-seven episodes. He didn't sit down and write a list. He just, slowly, dropped pieces of the show that felt like they belonged to someone else's podcast. Here is the list of what he dropped, in the rough order he dropped it, because the list is more useful than the prose.

1. 1.Drop the intro music. The intro music exists because radio shows had intro music, and radio shows had intro music because they needed listeners to know which show had just come on after the previous show. Your podcast does not have a previous show. The listener already chose to play your episode. They know which podcast it is. The intro music is a thirty-second tax on every episode that signals "this is a produced thing" before the produced thing has earned the production. Just start with the conversation. Helga's mid-sentence "…and the worst part is I actually liked the vets, you know?" is a better cold open than any synth jingle.
2. 2.Drop the show notes. Not entirely — you still need a description so podcast apps don't show a blank field. But drop the time-stamped chapters, the bulleted list of "topics covered," the carefully formatted links section. Replace it with one or two sentences in your own voice about what the episode is. Show notes that look like a corporate meeting agenda are signaling that the conversation inside is also a meeting agenda. Most listeners never read show notes anyway. The ones who do are reassured, not disappointed, when the notes feel handwritten.
3. 3.Drop the "before we start, hit subscribe" line. This is the single most format-aping move in indie podcasting and it makes every show worse. It interrupts the listener at the exact moment they were about to commit to the episode. It also signals that you, the host, are not confident the next sixty minutes will be good enough to earn the subscribe without you begging for it in advance. If the episode is good, they will subscribe at the end. If it isn't, no amount of front-loaded begging will save you. Cut the line.
4. 4.Drop the question list. Or at least, drop the version of the question list that lives in front of you during the recording. Tomás now writes a list of questions the day before, reads it three times, and then closes the document. He brings nothing into the room with him. The questions stay in his head if they matter and leave it if they don't. The conversation goes where the conversation goes. He hasn't asked a pre-written question on the air in eight months and the show has gotten better every episode.
5. 5.Drop the guest bio. You do not need to introduce your guest with their CV. Your guest is a human, not a LinkedIn profile. Skip the "Helga is the founder of a scheduling SaaS for Nordic hairdressers, she previously worked at" and instead let the guest introduce themselves through the conversation. Within ten minutes, the listener will know who Helga is in a way no bio can deliver. If the listener needs context up front, put it in the description, not in the audio. The audio is for talking.
6. 6.Drop the wrap-up. The "so where can people find you, plug your stuff, any final thoughts" segment is the most predictable two minutes of every interview podcast in existence. It is also where every podcast suddenly sounds like every other podcast, because the wrap-up format is industry-standard and identical across shows. End on the last real thing the guest said. If they said it twelve minutes before you stopped recording, edit out the twelve minutes. The end of the episode should be the end of the conversation, not the end of the format.

Each of these is a small thing. You could do any one of them and barely change the show. Doing all six together is what changed Tomás's show. Together, they make the podcast feel like a conversation instead of a deliverable. The listener can feel the difference without being able to name it. They will not write you a message that says "I love that you dropped the intro music." They will write you a message that says "this felt different, I don't know why." That message means you did it right.

What a small podcast actually is

I want to close with the part that took Tomás the longest to figure out, which is what a small podcast is actually for. Not what the gurus say it is for — which is "top of funnel" or "thought leadership" or "authority building," three phrases that should be illegal in any sentence about a Saturday closet podcast. What it's actually for, in the lived experience of someone making one.

A small podcast is one of the highest-bandwidth relationship tools an indie hacker has. If — and this is a big if — they treat it as a friendship engine rather than a content channel. Every episode is an excuse to spend ninety minutes in close conversation with another founder. Not at a conference, where you're standing up and someone keeps interrupting. Not in a Twitter DM, where the bandwidth is two hundred and eighty characters at a time. Not over coffee, where you both have to perform a little because there's a barista watching. In a closet, with mics on, where the only thing either of you has to do for the next ninety minutes is talk to each other.

Tomás now has, by my rough count, somewhere between thirty and forty indie founders he can text on any given Tuesday and get a real answer back from. Almost all of those relationships started in the closet. The podcast was the excuse. The conversation was the relationship. The eleven downloads were beside the point, because the point was never the downloads. The point was that he got to spend ninety minutes alone with another founder, and that founder is now a person who picks up when he calls.

If you treat the podcast as a content channel, you measure it by downloads, you optimize for downloads, you eventually quit when the downloads don't grow. If you treat it as a friendship engine, you measure it by whether the guest texts you a week later, you optimize for the kind of conversation that makes the guest want to text you a week later, and you keep going for fourteen months at eleven downloads because the conversations are doing the work even when the dashboard isn't.

Episode 28 was the moment Tomás's show became visible to people other than him and Helga and the six other listeners. But the show had been working since episode one, by the only measure that actually counted. He had twenty-seven new friendships. He just hadn't noticed, because he was looking at the wrong number.

Back to the closet

The reindeer blanket is still on the door. The used Shure is still doing the host mic. Tomás eventually replaced the worse Shure with a slightly better one, but only because the dent finally split. He records every other Saturday afternoon, except when he doesn't, in which case he records on a Thursday, or a Tuesday, or whenever the guest is in town. He no longer publishes on Wednesdays at 7 a.m. He publishes when the edit is done. Sometimes that's a Sunday.

His average episode now does between four and seven thousand downloads. Not because of episode 28, which has slowly become an outlier in his own back catalogue, but because the show that came after episode 28 was, finally, the show. He has a few episodes that have done more. He has a few that have done less. He has stopped checking the dashboard daily. He has not stopped reading the messages.

On Sunday mornings, when the editing is done and the next guest isn't booked yet and the apartment is quiet, he sometimes opens the podcasts directory and listens to twenty minutes of a show he's never heard before. Not the big ones. The small ones, the ones with fewer than five hundred episodes and a host who clearly recorded in their kitchen. He's looking for format ideas. Mostly he's looking for permission to keep dropping things from his own show. Every time he listens, he finds at least one small move he wants to steal. He calls it the quiet ritual. He says it's the only reason he hasn't gotten bored of his own podcast in three years.

If you're recording your second episode in a closet somewhere, and the download counter says four, and three of those are you — keep going. Drop the intro music. Drop the question list. Drop the apology you've been writing in your head for not having a bigger audience yet. Talk to a friend with the recorder on. Leave the pause in. The show is already working. You just haven't seen episode 28 yet.