A forty-minute video usually contains three or four genuinely self-contained moments. Not ten. The reason repurposing disappoints so consistently is that people go looking for ten, and the bar quietly drops somewhere around number five.
The only criterion that matters
A clip has to make complete sense to somebody who has no idea what came before it. That is a much stricter test than being interesting, and it is why the moment that landed best in the room is frequently the worst clip: it landed because of the twenty minutes of context in front of it, and that context is not coming with it.
What survives extraction is tension, a specific number, or a strong claim stated plainly. What does not survive is nuance, callbacks, and anything whose force depends on a relationship the viewer has not got.
Cut before the setup. The usable clip almost always starts several seconds after the speaker began the point, not where they began speaking.
The crop is an editorial decision
A 16:9 source cropped to 9:16 loses about two thirds of the frame. Which third you keep is a real choice, and it is one that automated tools make badly whenever there is more than one thing on screen. A conversation between two people has no correct crop, which is why the good clips from an interview are usually the ones where one person is talking uninterrupted.
Clipped audio is hard to follow
Speech pulled out of context is harder to parse than speech in context, and most viewers are watching on mute anyway. Captions on a repurposed clip are not an accessibility afterthought, they are the delivery mechanism, and the timing and line breaks are the whole visual craft of the format.
Podcasts and webinars are harder cases
A podcast has no visual component at all, so the captions have to carry everything and something on screen has to move. A static logo with text over it reads as an audiogram nobody asked for. A webinar is harder still, because it was paced for an audience that already agreed to spend an hour with you, and the prepared section is almost never usable.
In the webinar case the useful material is nearly always in the Q and A, because unscripted answers are the only part delivered at anything like short-form energy. Treat the recording as a source rather than a supply: most of it needs remaking rather than clipping.
The move that beats clipping
Take the claim rather than the clip. A single argument from a long video, remade as a purpose-built thirty-second vertical, outperforms an extracted version of the same argument almost every time, because it was written for the format instead of surviving it.
That reframes the whole exercise. The long video stops being an asset to be sliced and becomes research you already paid for, which is a considerably more productive way to think about it.