The direct answer
For a conversational podcast, plan on roughly one clip per five minutes of episode: three or four from a twenty-minute interview, eight to twelve from an hour, and rarely more than fifteen from anything, however long. That is the default sizing our pipeline applies when you turn an episode into clips, and it comes from watching what happens above and below it.
The number is a consequence of how many self-contained moments the episode actually contains. It is not a quota. Some forty-minute episodes have four moments that travel and some have eleven, and the honest set is whatever that count is.
Why more candidates make the set worse
It is tempting to ask for twenty clips and pick the best. Three things go wrong.
First, the extra candidates are not extra moments. Once the strong peaks are taken, the next candidates are second clips about the same peak, with worse boundaries, or moments from the negative list: intros, generalities, passages that depend on a visual. Reviewing them costs time and yields nothing.
Second, duplicates compete with each other. Two clips from the same story split the audience that would have shared one, and the feed learns that your channel repeats itself.
Third, the review step is where quality actually comes from, and review effort is finite. Twelve candidates get a real look each. Twenty-five get a glance. The set that ships is better when the set that is reviewed is smaller. How to judge each one is in what makes a good podcast clip.
Why fewer is also wrong
A floor matters too. An episode has an opening, a middle and a close, and a set of two clips almost always comes from the same ten minutes, which means the feed shows one side of the conversation. Three is the smallest set that can represent an episode, and it is the floor we use even for a short one.
How to decide for a specific episode
- Count the peaks first. Read the transcript and mark the sentences that would work as a quote. The method is in how to find the best moments in a podcast. The number of distinct peaks is your ceiling for this episode.
- Merge peaks that share a story. A story with one payoff is one clip, however many good lines it contains.
- Subtract what will not survive vertical. A demonstration on a whiteboard, a passage about a chart.
- Check the spread. If everything you have left comes from the first fifteen minutes, look again at the rest; the peaks are there, they are just quieter.
What remains is the set. If it is four, ship four.
Long episodes
A three-hour episode does not have thirty-six moments that travel. It has perhaps twelve to fifteen, spread unevenly, and the rest is the connective tissue that makes the long form worth listening to. Our pipeline caps a set at fifteen for this reason and selects in overlapping windows of a few minutes so that a quiet stretch late in the episode is not ignored because the first hour was loud.
Where the number lives in the workflow
When you run an episode through the podcast-to-clips workflow, the set is sized to the episode's length by default and every candidate arrives with a hook and a score, so the review is a matter of accepting, trimming or discarding. If you want a smaller set, discard; the pipeline will not have padded it with weak candidates in the first place. The wider workflow, from one episode to a week of posts, is in the complete guide to podcast clips.