Podcast to vertical reels: how to pick the moments that resonate

One podcast episode is 45 to 90 minutes of audio. Out of it you can cut six or more vertical reels that, added up, pull in more views than the episode did, and send the long-form-curious back to your main channel while they're at it. In the US, most people now find new shows through clip-reels rather than any other way: the Edison Research Infinite Dial 2025 puts it around 62% of podcast listeners, ahead of word-of-mouth at roughly 45% and app-store rankings at about 28%.

But "cut a podcast into reels" covers a huge range of quality. The same 60-minute episode can turn into eight clips with real viral potential or eight videos nobody finishes. What separates them is which moments you pick and how you dress them up. This is about which moments actually work, and which visual choices belong in the vertical format.

What counts as a good moment in a podcast

In a lecture the energy comes from the logic (see how to turn a lecture into reels). In a podcast it comes from the back-and-forth between hosts and guests, and that changes what makes a moment good:

Two positions clashing: a moment where the guest says one thing and the host pushes back. A short clip with that tension in it grabs people faster than any solo insight.
A reaction: one speaker reacting to the other, laughter, surprise, agreement, a flat no. It carries almost no information but plenty of feeling, and that's what holds a viewer.
A concrete story with a turn: the guest tells something short that flips at the end. "I thought X, turned out Y." It cuts cleanly to thirty or sixty seconds.
A three-line punchline: the host nails a concept in three tight sentences. Rare in real conversation, but when it happens it's a viral candidate. The algorithm should catch it by the verbal shape, an explicit "first, second, third" or "here's the deal: A, B, C".

What doesn't cut well: long ruminations with no clear payoff, technical bits that only make sense with the whole episode around them, third-party quotes (they come off as bare hearsay with no attribution).

In our experience about half the moments that look promising in an episode hold up against those four shapes. The rest fall short on one of them, or sit right on the line where the call comes down to who your audience is.

Dealing with more than one voice

The one real technical difference between cutting a podcast and cutting anything else long: you've got more than one voice to handle. Doing it well means three things working together:

Captions with every line tagged by speaker, either a label or a color, so the viewer always knows who's talking. Without that, dialogue captions turn to mush, especially when the guest and host talk over each other.
Reframing that keeps the active speaker prominent when two people share the frame in a face-to-face studio setup. A static crop on one of the two loses viewers.
A lower-third with name and role (host or guest) in the first three seconds. Without it the viewer has no idea who they're hearing.

Here's where the tools stand as of April 2026, honestly:

Opus Clip: automatic speaker labels, and generally the strongest on clean two-speaker studio recordings.
Vizard: also labels speakers automatically, and cuts more aggressively at the boundaries, so it often catches the reactions.
Lotima: no automatic speaker separation yet, though it's on the roadmap. Whisper transcribes the whole conversation as one track, and if a name or a term comes out wrong, you fix it once in chat and the fix sticks. For dialogue clips where per-speaker caption colors matter, Opus is the better fit right now. Where Lotima earns its keep on podcast material is picking a moment that carries a complete thought and showing you why, plus the montage layouts (split, picture-in-picture) for two people in frame.
CapCut Auto-Cut: no speaker separation at all, so it isn't really a podcast tool.

One thing about recording: put both people on a single camera mic in opposite halves of the frame and every tool downstream struggles, from the transcription to where the cuts land. Record it the studio way instead, a separate lavalier on each speaker and multi-track audio, and everything after it gets cleaner. If you're going to cut your podcast into reels regularly, that multi-track setup pays for itself within a few episodes in the time you save afterward.

What vertical podcast clips should look like

A plain talking-head reel doesn't work for a podcast. There's nothing for the eye to hold onto, and two heads in a split frame get tiring fast. What works:

A montage layout instead of a static two-shot: picture-in-picture or a split frame that keeps the active speaker large and the second person present, which gives the eye something to hold that a static crop can't.
Keywords burned into the captions, not every word, just two or three per thirty-second clip. Podcast audiences are used to dense information, and the highlights help them see which phrase carries the point.
A branded frame: a thin border with the show's logo, main color, and maybe the episode number. It's a marker for someone who's already seen two or three of your clips from different episodes and is starting to recognize the look.
A freeze on a reaction beat: if a moment has a strong reaction (laughter, surprise), you can hold a third to half a second of freeze-frame with an emoji over it. Don't overdo it, one or two per clip. More and it turns into loud TikTok-creator style, which reads as too much for podcast content.

The thing to avoid: full-screen, word-by-word captions in the viral-creator style. Podcast audiences skew older (the Edison 2025 median is around 38, against roughly 24 for native TikTok), and that treatment reads as unserious to them. Sentence-level captions with a keyword highlight hit the right balance of density and calm.

Where to post the clips

Podcast clips travel well to:

Instagram Reels, your main discovery channel. Thirty to sixty seconds is the sweet spot. Fifteen-second cuts from a podcast almost always lose viewers, since there's no time to land the thought.
TikTok, second by volume. The TikTok algorithm loves a clear emotional beat, laughter or surprise, and straight educational pieces underperform there.
YouTube Shorts, third by volume but first for turning viewers into subscribers, if your main channel is also on YouTube. Linking from a Short to the full episode (in the description or a pinned comment) gets us a 3 to 5% click-through in our customer data.
LinkedIn, which works well for B2B podcasts, the interview format with experts in management, tech, or finance. Keep these shorter, around 30 to 45 seconds, and always run a lower-third with the expert's credentials.
Threads and X, weak for video clips, since the format leans on text. If you post there at all, use it as a teaser that links out to the full clip.

For a solo podcaster, the smallest mix worth running is Instagram Reels plus one other platform, TikTok for a general-interest show, LinkedIn for B2B. Covering all of them is too much for one person. Six clips a week across two platforms is twelve posts a week, which you schedule through Buffer or Later in ten or fifteen minutes a sitting.

Why podcast scoring and lecture scoring pull apart

Back to the point from the lecture article: an AI tool that runs one moment-scoring model on both podcasts and lectures loses ground on whichever format it wasn't trained on.

You can see it plainly. Most tools in this long-to-short category, Opus and Vizard among them, trained on podcast data, so they lean toward interaction beats even on a lecture. But a lecture has no interaction beats, so the algorithm falls back on picking the loud moments by their sound, and that misses a lot.

It runs the other way too. A tool trained on lectures, and there are fewer of those, will skip the reaction shots on a podcast, because lectures don't have that signal and the model never learned to value it.

What this means for podcasters: if you're weighing Opus Clip, Vizard, and the newer arrivals like Lotima, then for a pure podcast Opus is still the strongest pick. For a mixed diet (podcast plus interview plus studio recordings plus lecture-style monologues), look at how the tool reasons about the source. Lotima reads the transcript's narrative rather than the loud spots and shows you why it picked a moment before it renders, and that reasoning carries across formats better than a scoring model tuned only for podcasts.

Talking to podcasters who put out one episode a week plus one lecture-style video, we keep running into the same setup: Opus for the podcast, manual cuts for the lecture, precisely because of this mismatch. It works, but it's two tools to keep up with.

A first pass, if you're just starting

If you record a podcast and want to try your first cuts:

Sort out the audio: make sure you have a separate track per speaker (multi-track export from Riverside, Squadcast, or Zencastr). If you recorded on one mic, do this week's episode with a lavalier on each speaker, and the jump in cut quality will be obvious right away.
Upload to the Opus Clip free trial (60 minutes a month free, enough for one test). You'll get twelve to fifteen candidates in ten to twenty minutes.
Run the candidates past the checklist: two positions clashing, a reaction, a concrete story with a turn, a three-line punchline. Out of twelve to fifteen you'll keep five to seven.
Polish: speaker labels, lower-thirds, the branded frame. If your base preset isn't set up yet, budget about half an hour for the first setup, then you reuse it.
Post on two platforms over a week, one reel a day. Check the numbers after seven days: which two worked best? That tells you which angle to push in the next episode.

That's about two hours total for the first week, then it settles around an hour and a half per episode. The other option, hiring a podcast-clipping editor at $20 to $50 an episode, only makes sense if you're recording more than two episodes a week or you genuinely can't stand review-style work.

Blog