Last month I timed something for my own curiosity. I took the last forty episodes that came through the studio, wrote down the runtime of the raw file and the runtime of the finished cut, and put the two columns next to each other. The average raw file was 71 minutes. The average finished episode was 48. That is 23 minutes of a host's life, per episode, that nobody ever hears.
None of those hosts set out to record 71 minutes. Every single one of them told me they were "going for about an hour".
So when someone asks me how long an episode should be, I have stopped giving the polite answer. The polite answer is "as long as it needs to be", which is true in the way that "eat healthy food" is true. It is correct and it helps nobody. Here is what I actually think after cutting a few thousand of these.
The honest answer is 25 to 45 minutes, and most people hate hearing it
For a show that is still growing, I think the sweet spot is 25 to 45 minutes, and closer to 25 than you want to believe. Not because attention spans are dying. That is a lazy thing people say. It is because a short episode forces a decision, and a long episode lets you avoid one.
When you have 90 minutes to fill, you never have to decide which of your five stories is the best one. You tell all five. Three of them are fine. One is great. One is the story you have told at four dinner parties and it does not work when it is not a dinner party. The listener gets all five and remembers none.
When you have 30 minutes, you have to pick the great one and build around it. That constraint is the entire craft.
I am not saying long episodes are bad. I am saying long episodes are earned. A three hour interview show works when the guest is genuinely so interesting that the length is the product. If you are twelve episodes in and your download graph looks like a heart monitor on a calm day, you have not earned three hours yet. Almost nobody has at episode twelve. I certainly had not.
Runtime is a symptom, not a decision
Here is the thing that took me too long to understand. Most hosts think of runtime as a target they set at the start. It is not. It is a number that falls out of choices you already made, mostly before you hit record.
The choices that actually set your runtime:
- How many questions you brought, and whether you were willing to skip any of them
- Whether you booked a guest who talks in paragraphs or in essays
- Whether you have a format, or you have a vibe
- Whether anyone in the room is allowed to say "let's move on"
- Whether you recorded a warm-up and then kept the warm-up in
If you fix those, the runtime fixes itself. If you do not, no editor on earth can save you, because we can only cut what you gave us. I can remove the dead air, the tangent about parking, and the four minutes where the guest looked for a document. I cannot remove the fact that the conversation never found its spine.
A 90 minute episode is not a long episode. It is usually a 40 minute episode with a 50 minute traffic jam in the middle.
What the drop-off actually looks like
I want to be careful here because I do not have your analytics and I am not going to invent numbers for you. But I can tell you the shape of the thing that hosts report back to me, again and again, once they start looking.
The listener does not quit at a random point. They quit at a specific one. It is almost always the same three places: about four minutes in, right after the intro, when the promise of the episode has not landed yet. Around the midpoint, when a tangent goes on too long and they realise they are no longer following anything. And at the point where the guest starts a story that is clearly going to take a while and there is no signal that it will be worth it.
None of those three are runtime problems. All three are structure problems that runtime makes worse.
Which is why I get slightly annoyed when the advice is "just make it shorter". Cutting a bad 70 minute episode to 50 gives you a shorter bad episode. You need to cut the right 20 minutes, and that is a judgment call, which is exactly why we watch every episode end to end here rather than letting a silence detector make the decision. A machine will find the pauses. It will not find the part where you already answered this question eleven minutes ago.
Solo shows and interview shows are not the same animal
If you record alone, everything above is more true and also harder. There is no second person to interrupt you, no natural handoff, no one whose reaction tells you the bit is over. The failure mode of a solo show is not boredom, it is drift. You start on one thing, it reminds you of another thing, and forty minutes later there is no through line.
I think solo episodes should be shorter than interviews by default. 15 to 25 minutes is plenty for a single point made well. I wrote out the whole approach in a post on keeping solo episodes tight, and most of it comes down to writing three bullet points before you record and refusing to leave them. If you already record alone and want someone to handle the assembly, that is essentially what our editing for solo shows work is.
Interviews are different. The runtime is partly not yours. It belongs to whoever you booked. Which means the skill is not planning, it is steering in real time, and steering means being willing to cut someone off. Politely, but genuinely. I put down the exact phrasings I use in a piece on how to interrupt a guest without being rude, because "so, on that note" is worth more to your runtime than any editing decision I can make afterwards.
The 90 minute episode that should have been two
Every few months an episode lands here that is genuinely two episodes wearing one trench coat. The first hour is one topic. The last forty minutes is a completely different topic that came up because the mic was still on and the conversation was going well.
That is a good problem. It is also a two-file problem, not a one-file problem. Splitting it costs you one recording session and gains you an extra publish slot, an extra title, and a second chance for someone to find your show. Our pricing works in your favour here too, since we bill up to 60 minutes as base and 90 minutes as more, so two 45 minute episodes are not a punishment. You can see the whole structure on the pricing section.
The catch is that both halves now need to stand alone, which mostly means both need a real title. A split episode with a title like "Part 2" is a split episode nobody clicks. I have a whole opinion about this in the episode title formula, and the short version is that the title should say what the listener gets, not what the topic was.
If you are publishing on YouTube as well, the same logic applies to the frame people see before they read anything. That is a different craft to editing and I do not pretend it is ours, which is why thumbnails go to our sister studio ClickTheory.
Recording quality changes what length is survivable
Here is an underrated factor. Good audio buys you runtime. Bad audio spends it.
If both people sound like they are in the same well-treated room, a listener will stay for an hour without noticing. If one side is thin, echoey, compressed by a video call, or drifting in and out, the same hour feels like ninety minutes and they leave at twelve. The tolerance for length is directly tied to how much effort listening takes.
This is the practical case for recording each person locally rather than capturing the call, which I explained in the piece on double-ender recording. It is more files to send us and it is worth every one of them. It is also the single cheapest upgrade available to most shows, because it costs nothing but a bit of setup discipline.
What I would actually do if I were starting again
If I were launching a show tomorrow, this is the plan I would follow, and I would not deviate for at least ten episodes.
- Target 30 minutes. Write it on the wall.
- Bring five questions, plan to use three, and treat the other two as spares.
- Record a hard stop. Tell the guest the hard stop before you start. Guests respect it and it changes how they answer.
- Cut the warm-up. Always. The first four minutes of any recording are two people finding out how the microphones sound.
- Publish weekly at 30 minutes instead of monthly at 90. Consistency compounds. Length does not.
- Let the show tell you when it is ready to be longer. It will. The signal is when you keep having to cut good material rather than filler.
That last one is the real test, and it is the only length rule I trust. If your edit hurts, you have earned more runtime. If your edit is easy, you were too long.
Where an editor actually helps
I want to be clear about what we can and cannot do with your runtime, because I would rather be useful than sound impressive.
We can take out the pauses, the false starts, the repeated question, the ten minutes where the topic went sideways, and the section that only worked because you were both in the room. We watch the whole thing to find those, which is why first cuts take 48 hours rather than four minutes. We will tell you when we think an episode should have been two, and we will tell you when we think you cut a good part out of nervousness. If you want to see how that back and forth runs, it is laid out in how it works.
What we cannot do is invent a shape that was never recorded. That part is yours, and it happens in the twenty minutes before you press record, not in the edit.
Pick a length. Defend it. Then let the show argue you out of it once it has something worth arguing with.
If you want a second opinion on whether your last episode was twenty minutes too long, send it over. I will watch it and tell you where I would have cut. samy@pixelsamy.com.