Most episodes do not fall apart because the host asked bad questions. They fall apart because the host asked good questions in the wrong order, spent nineteen minutes on the warm-up, and arrived at the reason they booked the guest with six minutes left.

A run of show fixes that, and it takes about twenty minutes to build.

This is for hosts who record interviews and have already decided that a full script sounds stilted. If you are still working out what to prepare before the guest arrives, start with preparing for a podcast interview and come back here for the clock.

What a podcast run of show actually is

A podcast run of show is a timed plan for one episode: the order of segments, how many minutes each one gets, and what has to happen before you can stop recording.

It is not a script. A script tells you what to say. A run of show tells you where you are.

That difference is the whole thing, and almost every template you will find online blurs it. Those documents are scripts with a duration column bolted on. You cannot glance at a script. You read it, and the moment you are reading, you have stopped listening to your guest.

It is also not your episode structure. Structure is the shape your show always has. The run of show is this Tuesday, with this guest, and the two questions you will regret not asking.

Radio worked this out about fifty years ago

Public radio calls it a clock, and it is not a metaphor. NPR shows run against a literal circular diagram dividing the hour into segments, and as the podcast 99% Invisible documented, the discipline around it is severe: segments cannot run long by even a second, because affiliate stations are automated to cut away exactly where the clock says they can. Directors talk about "hitting the post." Some of them have nightmares about missing it.

Here is the part worth stealing. In 2018 NPR redesigned the Morning Edition clock, a show that had been on air for nearly four decades. One thing the new clock was expected to fix was hosts cutting guests off mid-thought because they had run out of time. That was "something we have been struggling with under the old clock model," NPR's Sarah Gilbert told the network's public editor. "There wasn't quite enough room to finish that arc" of a conversation. The fix was structural: regroup the newscasts into two breaks and hand the lead block a full 11 minutes and 29 seconds of uninterrupted time.

They did not tell hosts to be better at interviewing. They changed the clock.

Your clock is doing the same thing to you right now, whether or not you have written it down. An unwritten run of show is still a run of show. It is just one you are discovering in real time.

The template

This is a run of show for a 45-minute published interview episode. Elapsed time, not wall-clock time, because your recorder shows elapsed.

Elapsed Segment Minutes Note
0:00 Levels, room tone, "we are rolling" 1 Never in the edit
0:01 Opening question, the easy one 4 First thing to cut
0:05 Theme A: why you booked them 14 Two starred questions live here
0:19 Theme B 12 Flexible
0:31 Theme C 9 First theme to compress
0:40 The close 8 Begins at 0:40 no matter what
0:48 Keep rolling, stop talking 2 The good stuff

Fifty minutes recorded for a forty-five minute episode. That gap is not waste, it is the edit's working room.

Four things make this document useful rather than decorative:

It is backtimed. Broadcast builds the clock from the end, not the start, because the end is the fixed point. Decide when the close begins and everything upstream has to fit behind it. Getting to the last ten minutes on time is the single highest-value thing a run of show does for you.

Themes get minutes, questions do not. Timing individual questions is how you end up steering a conversation into a wall. Themes are big enough to absorb a tangent.

Two questions per theme are starred. These are the ones you would be annoyed to have missed. Everything else in the theme is optional and you should feel nothing about dropping it.

Theme C is marked as the one that dies. More on that next, because it is the part everyone skips.

Decide what dies before you start recording

Something will run long. Not might.

The mistake is handling it in the moment, because in the moment you will take the time from whatever is in front of you, which is usually the section you are enjoying. Rank your themes by what survives a shortened episode, and write the ranking on the document.

Then the decision at minute thirty-four is not "what do I cut," which is a hard question asked under pressure while a person is talking to you. It is "we are eight minutes over, so Theme C goes," which is a lookup.

Thresholds worth setting in advance:

  • Four minutes over at the halfway mark. Ignore it. Interviews breathe, and you built a buffer for exactly this.
  • Eight minutes over. Drop the lowest-ranked theme entirely. Do not compress three themes by three minutes each, which is how you get an episode where nothing lands.
  • Fifteen minutes over and the guest is still going. You are not recording the episode you planned. Protect the close, get the starred questions asked, and sort the rest out in the edit.

The same logic rescues the opposite problem. When a guest shows up twenty minutes late, a ranked run of show is what lets you rebuild the episode in ninety seconds instead of improvising a shorter version of something you never finished planning.

Keeping it in front of you without reading it

A run of show you have to interpret while recording is worse than none, because interpreting costs attention you owe your guest.

One screen. No scrolling. If it does not fit, it is a script and you should cut it down.

Put the elapsed-time column on the left where your eye already goes, and set the starred questions in bold so peripheral vision can find them. Keep it on a second device rather than the one running your recording software, so a glance does not turn into a click on the wrong window.

And check it at the boundaries only. At the end of a theme, look down, confirm you are roughly where the clock says, look up. Three seconds. Checking it mid-answer is visible to your guest and it reads as exactly what it is, which is you waiting for them to finish. That impulse is also the thing most likely to turn a slow patch into an interview you are trying to rescue.

Where an AI producer helps

Building the clock is ten minutes of thinking and ten minutes of formatting. The formatting half is worth handing off.

Liyam AI builds a complete episode plan, with questions, visuals, notes and timing for every segment, either from your prep call (its agent listens in and drafts the outline, structures the segments and writes the questions) or from research you drop in as files, links and notes. You can also build the plan manually if you would rather own every line, which for a show with a settled format is often the right call.

What matters more is what happens once you hit record. Your run of show and your questions stay on screen while you record, organized by topic, so checking where you are costs a glance instead of a search. Liyam also generates follow-up questions in real time based on what is actually being said, which is the thing a static document cannot do. A written clock tells you a theme is running long. It cannot tell you that the guest just said something better than anything you prepared.

Use both. The clock protects the episode, the live prompts protect the conversation.

FAQ

What is the difference between a podcast run of show and a podcast script?

A script contains the words you will say. A run of show contains the order and the timing, and you glance at it rather than read it. Most hosts need a script only for the intro, the outro and any sponsor reads, and a run of show for everything in between.

How long should a podcast run of show be?

One screen, always. For a 45-minute interview that usually means six or seven rows plus a handful of starred questions. If yours runs to two pages, you have written an outline of the conversation rather than a plan for the recording.

Do you send the run of show to your guest?

No. Send them the themes and a few sample questions, which is a different document with a different job. Timings and your ranked cut list are production notes, and handing a guest a schedule tends to make them perform to it.

What should a podcast run of show include?

Elapsed time markers, named segments with a minute budget each, two starred must-ask questions per theme, and a ranked list of what gets cut first. Sponsor reads and any hard stops go in as their own rows.

Bottom line

Write the clock down, backtime it from the close, and decide in advance which theme dies when you run long. Twenty minutes of work, and it converts the worst moment in an interview, the one where you realize you are out of time, from a panic into a lookup.

NPR rebuilt a forty-year-old show's clock to stop hosts cutting guests off. You can do a version of that for Tuesday's episode, on one screen.

If you want the plan, the timings and the live prompts in one place while you record, Liyam AI is free to try.