15 to 20 clips from one interview
One long interview still yields 15–20 standalone clips. The economy is the cut list, captions, and platform length — not a second shoot.
One long interview still yields 15–20 standalone clips. That ratio is the 2026 repurposing economy, not a slogan: OpusClip's 2026 short-form trends puts it next to the length limits that actually matter — Shorts up to 3 minutes, Reels to 20 minutes, TikTok uploads to 10 minutes — and next to Shorts passing 70 billion daily views. Hedra and Pictory publish the same cluster (Hedra; Pictory). The interview is the expensive hour. The clips are how that hour gets paid.
This page is the cut list, not a model bake-off. Native-audio generators and interview clippers are different jobs. If the source is a conversation you already recorded, you are clipping, captioning, and packaging. You are not regenerating the guest.
The interview is inventory, not a single upload
A 45–60 minute conversation contains more standalone posts than anyone will watch as a single file. The mistake is uploading the long file everywhere and calling that distribution. The long file can live on YouTube or a podcast feed. The 15–20 are the feed products.
A standalone clip has to work if the viewer never hears the rest:
- A first sentence that is already a claim, a story beat, or a disagreement — not "that's a great question."
- A guest or host who is visually in the shot, or a caption that names them if they are not.
- A runtime that matches the idea, not the platform maximum. Most of these should be 20–45 seconds. A few can be 60–90. Almost none should be "we had three minutes of runway so we used three minutes."
- Burned-in captions, because a large share of the 70 billion daily Shorts views start muted.
What is not a clip: a highlight reel of seven quotes. That is a trailer for the long file. It does not stand alone, and it does not multiply. One idea per file. If two good sentences need each other, they are one clip, not two.
The 15–20 count assumes you recorded with clipping in mind. A single wide shot, bad audio, and no second camera will still yield clips. They will yield fewer that survive the first two seconds. Two frames (host, guest), a lav or a decent recorder, and a backup phone on the table is the production upgrade that actually changes the ratio. Lighting kits do not.
You do not need to fill every platform's new maximum. Shorts at 3 minutes, Reels at 20, TikTok at 10 are ceilings. Algorithms in that same 2026 cluster reward retention over polish. A 12-second clip that holds is the product. A 3-minute Short that exists because the cap moved is a different product, and it is usually the long interview uploaded again.
A cut list you can run in one sitting
Do not open an editor and "see what you find." Open a transcript. Mark the lines. Then cut.
Pass 1 — claims. Any sentence that could be the first line of a post. Specific number, specific disagreement, specific story beat. Mark the timestamp. You want more marks than you will ship; the kill list comes later.
Pass 2 — pairs. Some claims only work with the previous sentence ("we tried X" needs "it failed because Y"). Circle those as one clip. Do not split them to inflate the count.
Pass 3 — dead air. Intros, ads, "before we get into it," tech checks, the last two minutes of goodbyes. These are not clips and they are not B-roll. Delete them from the candidate list.
Pass 4 — rank. Fifteen to twenty is a shipping target, not a scraping target. Rank on: would this line stop a stranger, is the picture usable, is the audio clean enough to caption. Ship the top of the list. Keep five as a bench for the week the first ten decay.
That sitting is the "per hour" in the slug. A recorded hour, plus a focused hour on the transcript, is the unit. Teams that report up to 11× more video per month with AI workflows (OpusClip / Hedra / Pictory) are not talking about 11× more interviews. They are talking about this ratio, plus captions, plus batch export.
Tools that auto-suggest clips are a first pass, not a cut list. They over-index on energy and laughter. Energy is not a claim. Watch the suggestions with the sound off. If the caption would be "um, so yeah," skip it even if the waveform is loud.
For each keeper, write the on-screen title in the same pass. The title is often the hook. If you cannot title it in a sentence the guest actually said, it is not a clip yet.
Package for three lengths, not three personalities
The same quote should not be three different performances. It can be three different packages:
- Short (under ~45s). One claim, captions, a name slate if needed, hard out. This is the default for Shorts, TikTok, and Reels.
- Medium (1–3 min). The claim plus the story that earned it. This is the Shorts-maximum case, or a Reels cut for people who will sit. Do not pad.
- Long. The interview, on the long-form channel, with the clips pointing back if you want them to.
You rarely need a 10-minute TikTok of the same interview you already posted as a podcast. The 10-minute cap is useful when the clip is a ten-minute idea (a walkthrough, a live). It is not a reason to stop clipping.
Captions are the package. Run them as a batch on the keepers. The caption generator and captions for silent autoplay are the finishing path. One style across the 15–20 so the series looks like a series. Safe area: keep the hook line out of platform UI. If the interview already has burnt-in Zoom names, recrop or cover them; two caption systems on one frame is how clips look unfinished.
A hook line on screen can be the guest's sentence or a tighter rewrite of it. Do not write a clickbait title the guest did not earn. The retention those trend posts describe is hold, not bait-and-switch.
Do not generate a new guest. If you want B-roll, generate or pull cutaways that do not impersonate the person. A cutaway of the product they named is legitimate. A generated talking head of the guest is a disclosure and likeness problem, and it is not "repurposing."
What still does not scale
The interview still has to be good. Fifteen clips of a vague conversation are fifteen files of nothing. Book guests for sentences, not for logos. Ask questions that force a specific: "what did you try that failed," "what number changed," "what would you not do again." The clip ratio is a function of the answers.
The person still has to:
- Choose the 15–20, which is a taste job.
- Kill the twins (three clips that are the same claim in different ums).
- Write the first caption line.
- Decide which clip belongs on which platform that week, so you are not dumping twenty files into one account in an hour and teaching the system you are a farm.
- Live with the fact that some of the twenty will get no distribution. The ratio is a production ratio, not a guarantee of 20 hits.
YouTube's inauthentic-content policy and Instagram's unoriginal-content rules are the ceiling on "just clip everything." A labelled, specific clip from an interview you hosted is yours. A channel that is only other people's interviews, templated, is a different risk. The 15–20 economy assumes the conversation is yours to cut.
If the source is not an interview — a keynote, a demo, a live — the same math holds with a tighter kill list. Demos clip into features, not vibes. Keynotes clip into claims. Lives clip into moments that survive without the chat.
The AI video generator is the wrong first tool for this job unless you need a cover frame or a cutaway. Start with the transcript and the caption rail. Sign up if the missing piece is captioning the twenty in one pass rather than in twenty tabs.
FAQ
Does every interview actually produce 15–20 clips?
Only if you recorded claims, not vibes, and you are willing to ship short files. A 20-minute chat with no specific will produce three clips and a lot of wishful marking. The OpusClip 2026 figure is a working ratio for a real conversation, not a quota.
Should I post all 20 the same day?
No. That is how a series looks like a dump. Ship a few, hold a bench, let creative fatigue on the channel tell you when the next claim should go. The production day can be one sitting. The publishing calendar should not be.
Is a 3-minute Short better than a 30-second clip because the cap moved?
Not by default. Retention over polish is the other half of that 2026 cluster. Use the longer cap when the idea needs the time. Use 30 seconds when it does not. The cap is permission, not a brief.