Podcast clips for social media: which moments to cut

Three tests a clip has to pass

Most weak podcast clips are not badly edited. They were cut from the wrong place. A listener who loved the full episode will forgive a clip that needs context, but a stranger scrolling a feed will not. Run each candidate through these three tests before you caption anything.

Length follows from the ideas. Most clips that pass land between 20 and 90 seconds. KeyMo's length settings are Short at 20 to 40 seconds, Medium at 40 to 60, Long at 60 to 90, and Auto at 15 to 90. If a thought needs two minutes, it probably contains two thoughts.

Three worked examples from different shows

These are illustrative cases, not real episodes, written to show where the cut goes.

• Founder interview. The guest tells how a launch failed. The host's question ('what went wrong?') is dead air for a stranger, so start on the guest's answer: 'We shipped on a Friday and by Monday nobody had logged in.' Stop after the lesson, before the next question. A 45-second Medium clip. The payoff is the lesson, so end on it.
• Two-host banter show. The funniest minute is made of three callbacks to earlier jokes. A stranger will miss them. Look instead for a self-contained bit, a rant or a story with its own set-up. Start on the first line of the story, end on the punchline, and cut the laughter that follows. Short, around 25 seconds.
• Expert Q&A. A listener asks how to pick a running shoe and the guest gives a five-part answer. A single clip of the whole answer is too long and loses people at part three. Make the first rule a clip on its own, with the question as on-screen text, and keep the other parts for the next days. Long, up to 75 seconds, or two Medium clips.

The pattern: start on the answer, not the question, cut off the follow-up, and treat repeated jokes or references as a reason to look elsewhere.

What platforms say about length and the first seconds

No page promises that a clip will do well, and nobody can honestly promise that a clip will spread. What the platforms do publish is narrower and more useful.

• Instagram for Creators says it recommends videos of three minutes or less to people who do not follow you yet, and advises making the first three seconds engaging. Instagram's Help Center says reels over three minutes are not recommended to new audiences.
• YouTube describes Shorts reporting around 'stayed to watch', the share of times viewers stayed past the initial seconds. It does not state how many seconds.
• TikTok's advertising guidance recommends the hook in the first 6 seconds and the main proposition in the first 3. That is ad advice, but it reflects the same pressure on organic posts.
• YouTube accepts square or vertical Shorts up to three minutes.

So the first line matters more than the length. See the first 3 seconds guide for how to write it.

Podcast Reels and Shorts ideas that keep working

If you are staring at a transcript with no idea what to cut, look for these moment types. They stand alone because each carries its own setup.

• The hot take: one clear opinion, then the reason for it.
• The story with a turn: three sentences of setup, one of surprise.
• The direct answer to a listener question, with the question shown as text.
• The disagreement: two people differ in a way you can follow without the episode.
• The correction: 'I used to think X. Then Y happened.'
• The definition: a term explained simply, which is good for search as well as feeds.

The best podcast clips tend to be the ones a listener would forward with 'listen to this bit'. The best podcast Reels and Shorts are the same clips, reframed to vertical and captioned.

Where KeyMo fits in the picking

You can run the three tests by reading a transcript, but doing that for a 60-minute episode is the slow part. KeyMo does the first pass. It scores each moment out of 10 with a one-line reason, drops anything below 7, and gives you up to 15 captioned 9:16 clips as a shortlist. Built-in targets are 3 to 5 clips from a video up to 15 minutes, 5 to 8 up to 30, and 8 to 12 up to 60.

Those are targets, not measured results, and the final call is still yours: read each reason, watch the first three seconds, and use the chat editor to move a start point. Start with the podcast clip generator, or read how to clip a podcast for the full steps.