Last month I counted the revision notes on one week of episodes. Forty one notes across five shows. Thirty of them were one line each and took under two minutes to action. Six were real editorial calls. Five were the same person asking for something they had already asked for and then unasked for two episodes ago.
That last group is the whole problem with revision rounds. Not the work. The churn.
I have been editing podcasts long enough to have opinions about this, and here is mine: the revision round is not a quality checkpoint, it is a communication format, and most shows are using a format that guarantees a second round. Fix the format and the second round mostly disappears. That is not a productivity hack, it is just what happens when notes stop being ambiguous.
The round that actually works is one pass, timestamped, in one place
Here is the version I push every host toward.
You get the first cut in 48 hours. You watch it once, start to finish, without pausing to write. Then you watch it again and write notes as you go, every note anchored to a timecode, all of them in a single document or a single message. You send that once. Revisions come back the next working day.
That is it. One pass, timestamped, in one place, sent once.
It sounds obvious written down. Almost nobody does it. What actually arrives is four WhatsApp messages over two days, one of which says "the middle bit feels off", followed by a voice note at midnight saying to ignore the second message.
Every one of those is a request for a guess. And a guess is what generates round two.
Watch it once before you write anything
The single highest leverage habit I can give a host costs eight extra minutes of their life.
Watch the cut all the way through with your hands off the keyboard.
The reason is that most first-listen reactions are not edit notes, they are pacing sensations. You feel something is slow at minute nine, so you write "tighten minute nine". But the actual problem is that the setup at minute six was cut too tight, so minute nine has no context to land against and it drags. If you had watched the whole thing first you would have felt where the air went missing.
Editors can only fix what you point at. Point at the symptom and you get the symptom treated. Then you watch round two, still feel it, and write "still feels slow" which is the least actionable note in the English language.
Notes written during the first watch describe how you feel. Notes written during the second watch describe what to do. Only one of those is useful to the person holding the timeline.
Timecodes are not bureaucracy, they are the whole message
"Cut the tangent about the pricing thing" sounds specific. It is not. In a 55 minute episode there may be three moments that qualify. I will pick one. There is a decent chance it is not yours.
"14:22 to 15:40, cut the whole tangent" takes four seconds longer to write and removes the guess entirely.
The rule I give people is simple. If a note does not have a number in it, it is not a note, it is a feeling. Feelings are welcome, they just belong in a different part of the document.
A note format that works:
- In and out points for anything to remove, as
mm:ss to mm:ss - A single timecode for anything to add, fix, or change, as
mm:ss - One sentence on what you want, in plain words, not editing vocabulary
- A separate short section at the bottom for whole-episode feelings, clearly marked as such
That last bullet matters more than it looks. Hosts have real instincts about the shape of their show. Those instincts are worth hearing. They just cannot live inside the timestamped list, because a global feeling attached to a specific timecode makes an editor change one thing when the show needed a different approach across ten.
Every few weeks someone writes "add a J cut here" and means something completely different from what a J cut is. Or asks for "a jump cut" when they want a hard cut with no visual break, which is nearly the opposite.
Say what you want in normal language. "I want the next speaker's voice to start before we see them." That is unambiguous. It takes zero technical knowledge and it is impossible to get wrong.
This applies with force on multi-cam. Camera direction is where vocabulary confusion turns into two extra rounds. "Use the wide more" could mean more frequently, or for longer, or specifically during laughs. On multi-cam editing I would rather have "hold the wide through the whole laugh, do not cut back until it dies down" than any amount of correct terminology. If you are running multiple angles and finding the review painful, it is often not the notes at all, it is the media. Choppy scrubbing on a review link makes people write vague notes because they cannot land on a frame, which is one of the quieter arguments for proxy workflows on multi-cam shows.
Separate "wrong" from "different"
This is the distinction that decides whether your revision round is one round or three.
Some notes are corrections. A name is misspelled in a lower third. The audio drops for two seconds at 31:10. The intro music is three seconds too loud under the first line. These are objectively wrong and they should always be fixed.
Other notes are preferences. You would have held the shot longer. You would have kept the joke. You want the chapter break somewhere else. These are legitimate and they are also negotiable.
Mixing them in one undifferentiated list means everything gets treated with equal weight, and preference notes quietly multiply because there is no natural end to them. Mark corrections as corrections. Then you can look at your preference list, see fourteen items, and decide honestly that six of them are you being a perfectionist about a conversational show nobody will rewatch frame by frame.
Corrections are also where a technical problem is hiding in plain sight sometimes. If the same kind of sync note keeps appearing at roughly the same point in every episode, that is not an editing preference, that is a recording issue, and it is worth reading what to do when audio desyncs before writing the same note a fourth time.
Decide what a revision round is for before you buy one
Our Starter tier at $199 per episode includes one revision. Studio at $299 includes unlimited revisions.
I am not going to pretend unlimited is always the better buy. It is the right call for shows where the format is still moving, where you are testing graphics, changing cold open structure, figuring out how much b-roll you actually want. It is genuinely useful to iterate there.
But if your show is settled and your notes are timestamped, one revision is usually enough, and the money is better spent on things the audience actually sees. A vertical clip at $25 or show notes and chapters at $35 do more for a stable show than a fourth pass on a cut nobody but you will scrutinize. I would rather you spend on distribution than on polish you cannot perceive on a phone speaker.
There is a real fork here between templated and custom graphics too. Custom graphics generate more revision notes by design, because there is more to have an opinion about, which is the honest tradeoff in templates versus custom podcast graphics.
The rounds you should not be spending on
Here is where I will be blunt. A meaningful share of revision notes are about the wrong artifact.
Hosts scrutinize the cut and ignore the packaging. Then the episode does not perform, and the conclusion is that the edit needed another pass. Almost never true. On YouTube the thumbnail and title decide whether anyone reaches your edit at all, and the platform difference is real enough that it is worth thinking about where the video version actually lives before optimizing frames. If the packaging is the weak link, our sister studio ClickTheory does thumbnails at 8 for $250, and that budget will beat a fourth revision round every single time.
The same logic applies to audio quality, chapter placement, and clip selection. All of those move numbers. Trimming another 400 milliseconds off a pause does not.
Where automated tools quietly cost you a round
I have a bias here and I will state it plainly. We do not auto-cut by silence detection. A human editor watches the whole episode.
The reason is directly relevant to revisions. Silence detection removes pauses without knowing which pauses were doing work. A two second beat after a hard question is not dead air, it is the moment the answer lands in. Cut it and the interview reads as rushed, and the host feels it without being able to name it, and you get "the middle feels off" instead of a timecode.
That is a whole revision round created by a tool that was supposed to save time. I go into more of the tradeoff on Descript versus a human podcast editor, because there are real cases where the automated route is fine and I do not think pretending otherwise helps anyone.
A quick sanity check before you hit send
Before you send your notes, read them back once and ask three things.
- Does every actionable note have a number in it
- Have I marked what is a correction and what is a preference
- Am I contradicting something I asked for on a previous episode
That third one is the sneaky one. Shows drift. You asked for tighter cold opens in March and looser ones in June and never told anyone that you changed your mind. Say it out loud. "I have changed my mind about cold opens, here is the new default." One sentence, and it stops being a per-episode note forever.
Standing preferences belong in a show bible, not in every revision round. If you are telling us the same thing every week, that is a process failure on our side as much as yours, and the fix is to write it down once.
One watch to feel it. One watch to write it. Timecodes on everything actionable. Corrections separated from preferences. Sent once, in one place. Standing preferences promoted out of the round and into the defaults.
Shows that work this way get their episode finished in a day after the first cut. Shows that do not can spend a week on the same episode and end up somewhere very close to where the first cut already was.
None of this is about being a low maintenance client. Ask for what you want. Ask for a lot of it. Just ask for it once, with a number attached, and you will get it faster and you will get it right.
If you want to see how the handoff and first cut timing actually run, how it works lays out the whole sequence, and I am at samy@pixelsamy.com if you want to argue with any of this.