The Long-Video Clip Factory

The problem

You finished the Zoom call Tuesday afternoon — a 47-minute coaching conversation with a client who let you record it for your own library. You stopped the recording, said you would “make some clips later,” and dropped the file into a folder on your laptop called recordings/2026/. It sits there now next to forty-two other recordings just like it. Webinars, podcast interviews, cohort sessions, internal workshops. Two hundred plus hours of long video. Maybe ten of them have ever produced a short-form clip. The other thirty-two have collected dust because the second-time-around effort to extract clips — scrubbing the timeline at 0.5x speed, finding the moment, exporting the cut, captioning it, sizing it for each platform — looks like another half-day per video. So it never happens.

Hidden inside each 45-minute recording are somewhere between ten and fifteen publishable short-form clips. Standalone insights, counter-intuitive claims, numbered lists, stories with a point. Across forty-two recordings, that is a library of roughly four hundred unpublished clips sitting on your drive while you spend Tuesdays recording yet another long video that will collect 240 views and disappear. Every creator I have worked with has this same drive. The math is brutal once you do it: a creator who publishes one long video per week plus four to six platform-specific clips from it grows roughly four times faster on every measurable surface — same source material, same recording sessions, only difference is whether the clips got made.

What most people get wrong

They scrub the timeline looking for clips. This is the workflow that kills repurposing dead before it ever becomes habit. You open the editor, you play through at 1x or 0.5x, you scroll back, you note a timestamp on a sticky, you scrub more, you find maybe two more moments in twenty-five minutes of attention. By the end of an hour you have three candidates and your brain is fried. A round trip on that workflow is four to eight hours per video to produce three to five clips — sixty to ninety minutes per clip. The math kills it before it ever reaches a weekly rhythm. The fix is not “scrub faster” or “use a better editor.” The fix is to stop searching for moments in a linear medium and start searching for them in text. A 45-minute transcript is six to nine thousand words; you can scan it for clip-worthy patterns in twelve minutes and surface fifteen to twenty candidates. Same source material, one-fifth the time, four times the candidate pool.

They cross-post one file to four platforms. The temptation is real — you cut a strong 50-second clip, you have it sized 9:16, why not just upload the same MP4 to YouTube Shorts, TikTok, Reels, and LinkedIn and call it shipped? You can. The result will be 60 to 80 percent under-performance on each platform versus the variant approach. The platforms detect cross-posted content (watermarks, frame-rate signatures, caption styles) and de-prioritize it in feeds. The viewers can also tell — a TikTok that looks like a LinkedIn clip gets scrolled past in two seconds because the visual grammar is wrong. The four platforms have meaningfully different rules: TikTok is sound-on by default and punishes polished captions, LinkedIn is muted-by-default and rewards polished captions, YouTube Shorts caps at 60 seconds, Reels rewards trending audio under your voice. Ignoring those rules costs you most of the reach you were trying to capture.

They publish in bursts and burn out. A creator gets motivated on a Sunday, posts ten clips across four platforms in two days, runs out of source material, then publishes nothing for three weeks until the next motivation cycle. The algorithm punishes inconsistency. A creator who posts fourteen pieces evenly distributed across seven days outperforms a creator who posts fourteen in a single afternoon by roughly three to five times. The whole game is cadence — calendar-locked, scheduled, batched in advance, then dripped out. Most creators who fail at repurposing fail at the calendar discipline, not at the editing.

A working approach

The book is organized as five stages of a single workflow. Each stage compresses the time per clip by removing a category of friction.

STAGE 1 — Audit the video library (90 min, one time)
  Score each long video on topic density, audio, visual interest
  Decide: must-repurpose / archive / delete
  Output: a prioritized queue, usually 8-15 videos

STAGE 2 — Transcript-first clip discovery (30 min per long video)
  Generate transcript with Descript / Whisper.cpp / Otter
  Scan for 6 patterns: standalone insight, counter-intuitive claim,
  numbered list, story with point, strong opinion, direct Q&A answer
  Output: 15-25 highlighted candidates

STAGE 3 — Score against Hook-Point-Payoff (15 min)
  Hook quality 1-5 + Payoff quality 1-5 + first-three-words test
  Cut preamble aggressively (8-15 sec off the front of every candidate)
  Output: 10-15 publish-ready candidates

STAGE 4 — Platform-specific variants (40 min per clip, batched)
  YT Shorts 60s / 9:16 / muted preview / karaoke captions
  TikTok 60-90s / 9:16 / sound-on / native casual captions
  Reels 60-90s / 9:16 / muted preview / trending audio underneath
  LinkedIn 60-180s / 1:1 or 9:16 / polished captions / 3K-char post
  Output: 4 variants per clip, styled native to each platform

STAGE 5 — Batch caption, schedule, drip (2 hours Saturday)
  Submagic batch styling with platform presets
  Audio cleanup via Descript Studio Sound or Auphonic
  Schedule via Buffer / Later / Metricool to a 14-day drip
  Output: 14 pieces of distribution per week

The library audit is the single most valuable hour you will spend on the entire workflow because everything downstream gets prioritized by it. Open a spreadsheet — the bonus folder includes clip-scoring-spreadsheet.csv with the columns pre-populated — and list every long video you have published or recorded in the last 24 months. For each one, score it on topic density (how many distinct standalone ideas does it contain), audio quality (5 is studio mic with no echo, 1 is phone speakerphone), and visual interest (5 is animated face on camera, 1 is audio-only with a static cover). Add a time-sensitivity tag: evergreen / dated / outdated. A typical audit produces three buckets: 8-15 must-repurpose videos at the top of the scoring, 15-25 in the archive bucket, and a dozen that should just be deleted because they reference deprecated tools or contain consent issues from old client calls. Be ruthless on the deletes. A bloated archive is a procrastination magnet — every time you open the folder, the bad videos drain attention away from the good ones.

Transcript-first is the single most important shift

This is the inversion that compresses per-clip cost from sixty minutes to fifteen. Stop looking for clips in the video timeline. Look for them in the transcript. Video timelines are linear — to find a moment you have to play through, to verify it you have to rewind, to compare two candidates you have to load both. The cognitive load is high, the search is slow, the dropoff is brutal. Text is parallel — you scan eight hundred words in 90 seconds, pattern-match the quotable moments on the first read, and copy candidates into a list. Seven times faster on average, measured across timing tests on the same source video.

Three transcript tools cover the range. Descript Pro at $24/month is the right default for most creators because the transcript lives inside the editor where you will eventually cut the clips — the roundtrip from “find the moment in text” to “select the words to clip” to “export the cut” is roughly 90 seconds. Whisper.cpp is the free path: same model that powers most commercial services, runs locally on your laptop with one command, takes 5-12 minutes to transcribe a 45-minute video. The catch is you then have to manually align timestamps to the video, which Descript does automatically. Otter.ai at $17/month is the middle ground if you want a standalone transcript with speaker labels but do not need the in-editor cut workflow. Riverside.fm includes transcription if you happen to already record there.

Read the transcript once, top to bottom, scanning for six specific patterns. The standalone insight (a 30-60 second passage that stands alone — a claim plus the reasoning, complete in itself). The counter-intuitive claim (“most people think X, but actually Y”). The numbered list (“there are three reasons…”). The story with a point (60-90 seconds of narrative ending in a named lesson). The strong opinion (a clear position on something contestable). The direct answer to a Q&A question (the question becomes the hook, the answer becomes the payoff). Highlight in two colors — yellow for “definitely going to clip” and green for “maybe.” Most 45-minute transcripts produce fifteen to twenty-five highlights of 30-90 seconds each. That is your candidate pool.

The Hook-Point-Payoff formula

Every clip that works on any platform is three beats. The hook earns the first three seconds. The point earns the next thirty. The payoff earns the watch-through to the end. Miss any one and the clip dies — not because the content was bad but because the algorithm measures completion percentage as the dominant ranking signal, and the three drop-off points in short-form viewing happen at exactly those three checkpoints.

The hook (0-3 seconds) is the entire game. A clip that loses 60 percent of viewers in the first three seconds will not be shown to more people regardless of how good the rest is. Four hook patterns consistently work: the contrarian claim (“most people think X, but they’re wrong”), the specific number (“87 percent of new founders make this mistake”), the provocative question (“why do you think your sales calls don’t close?”), and the promise of a reveal (“I’m going to tell you the one thing that took me five years to learn”). What does not work: throat-clearing. “Hey everyone, so today I wanted to talk about…” loses 70 percent of viewers before the sentence finishes. Cut every clip to start mid-thought, with the verbal hook intact and the lead-in stripped.

The point (3-30 seconds) is the central claim — the answer to “what is this clip about, in one sentence.” If a viewer fifteen seconds in cannot tell you what the clip is going to be about, they are leaving. The point usually arrives 4-8 seconds in and is then unpacked across the next twenty-some seconds. The clip is functionally over by second thirty. The remaining time is the payoff that justifies the watch-through.

The payoff (30-60 seconds) is what the viewer remembers and what they share. Three forms work consistently: the concrete example (the point was the claim, the payoff is the specific instance — “I had a client with a 12-person SaaS who tripled prices and lost four customers but tripled net revenue”), the tactical step (the point was the principle, the payoff is the actionable thing they can do tomorrow morning — “write your subject line last, after the body, and make the first four words name the specific benefit”), or the reframe (the point was the conventional view, the payoff is the new lens — “hiring is about pace, not skills; skills can be taught, pace is built into who someone is at 25 and won’t change”).

Before you commit to clipping any candidate, apply the first-three-words test. Read the proposed opening out loud. “So basically what…” — cut, throat-clearing. “I think that…” — cut, hedging. “Hey everyone, today…” — cut hard, lowest-performing opening on any platform. “Most people think…” — keep, strong contrarian hook. “87 percent of…” — keep, specific number, immediate authority. Most clip candidates lose 8-15 seconds of preamble that should be trimmed off the front. The difference between a clip that dies in three seconds and one that survives the first checkpoint is almost entirely that aggressive front trim.

Platform-specific variants

A single 50-second clip exported as “one video for all platforms” will under-perform on every platform compared to four platform-specific variants. The 25-40 minutes per clip you spend on the variants buys you 8-12x the combined reach. The book ships twenty variant templates — five per platform — in clip-variant-templates.md in the bonus folder. The architecture of each platform:

YouTube Shorts caps at 60 seconds hard (longer uploads route to long-form), aspect ratio 9:16, default playback muted in feed previews. The platform rewards completion percentage above almost any other metric — a 30-second clip holding 90 percent of viewers outperforms a 55-second clip holding 70 percent, even though the longer clip technically delivers more watch-time. Implication: when in doubt, cut shorter. Hook tolerance is the longest of the four platforms — you have 4-5 seconds before drop-off becomes punishing. Caption style is high-contrast karaoke-style word-by-word animation, 40pt+ at 1080x1920. Title formula: hook + specific + searchable keyword in 60-80 characters. Always include #Shorts in the description plus two or three topic tags.

TikTok runs 60 seconds for new accounts and 90 seconds+ once you build a follower base, aspect ratio 9:16, default playback sound-on and vertical fullscreen. This is the sound-on platform — the verbal hook matters more than on any other surface because viewers actively hear the first two seconds before deciding to scroll. Hook tolerance is the shortest: 2-3 seconds before For You page viewers swipe past. TikTok punishes “corporate” or “polished” video styles — captions need to feel native to TikTok, not lifted from a slide deck. Submagic ships a TikTok preset that mimics the platform’s own bouncy, slightly-imperfect caption style. The caption text under the video uses the 2,200-character budget for context plus 3-5 hashtags mixing specific and broad. Cover image: choose a specific frame, not the auto-generated thumbnail.

Instagram Reels allows up to 90 seconds with 60 seconds the recommended sweet spot, aspect ratio 9:16, default playback muted in feed previews. Reels rewards remix culture — trending audio under your voice at 5-10 percent volume so the words still come through. The clips that perform best use trending audio layered with the speaker’s voice, supported by Submagic and CapCut native. Hook tolerance is 3 seconds, similar to Shorts, but the first frame must be visually arresting because Instagram feed scroll is muscle-memory fast. Caption style is sentence-case white text with thin black outline, no all-caps. Hashtags cap at 3-5 native tags — not the 30 you might use on legacy Instagram posts.

LinkedIn runs 10 seconds to 10 minutes with the engagement sweet spot at 60-180 seconds, aspect ratio 1:1 square or 9:16 vertical (square performs better on desktop), default playback muted. LinkedIn is genuinely different from the other three. It is a sound-off platform by default (people scroll in meetings and on commutes), so captions are doing 90 percent of the work. It is a longer-attention-span platform — the 90-second clip that gets skipped on TikTok routinely holds attention on LinkedIn. And it is the only platform where the post-text caption is more important than the video — LinkedIn viewers often read the post text first and only watch the video if the caption hooked them. Use the full 3,000-character post-text budget for context plus takeaways plus CTA. Hook tolerance is the most forgiving of the four — 5-6 seconds — and the extra time should establish business credibility (your role, your specific result, your specific number) before the central claim. “I ran a 240-person sales org for 4 years. Here’s the one thing” is a stronger LinkedIn opening than a generic contrarian claim. Hashtags: 3-5 professional tags, #Leadership not #fyp. Click-through rate from LinkedIn short-form to long-form sources runs around 2.8 percent — roughly seven times the equivalent rate on TikTok or YouTube Shorts.

Batch captioning and scheduling

Per-clip editing kills the whole workflow. If you batch-edit one clip at a time, the per-clip cost stays around 30-40 minutes. At that rate, 12 clips per week is a 6-8 hour weekly commitment that most creators cannot sustain. The fix is the 2-hour Saturday morning session: load all the week’s source clips, apply platform presets across all of them, batch-export, schedule. The per-clip cost in batch mode drops to 12 minutes — a 3x compression over single-clip mode.

The tool combination most worth the money: Submagic at $20/month for batch caption styling with platform presets (it ships YT Shorts, TikTok, Reels, and LinkedIn presets that match each platform’s caption aesthetic), plus Descript Pro at $24/month for the underlying cut and the Studio Sound audio cleanup. That is $44/month — the minimum viable paid stack. Opus.pro at $19/month is the alternative path: it takes the long video as input and auto-proposes 10-20 clip candidates with timestamps, which can cut the highlighting pass time in half. Its weakness is that the AI suggestions are 50-70 percent on target, so you still review and reject. The combined workflow that works for many creators: Opus for clip discovery, then Submagic for the final styling. Captions (app) at $10/month is the mobile-first alternative for on-the-go editing. CapCut Desktop is free with no watermark and no time limits — the workflow takes 30-50 percent longer per clip than the paid stack because there are no platform presets shipped and no batch operations, but for creators producing under five clips per week the time difference is manageable. For audio cleanup beyond Descript, Auphonic at $24-48/month depending on volume runs in batch and auto-normalizes levels, removes background noise, and improves voice clarity.

The Saturday session structure: load all 12 source clips at the 0-15 minute mark, style them across all four platforms from 15-75 minutes (the first clip takes 25 minutes because the preset needs setup; clips 5-12 hit the 10-minute-per-clip target), generate thumbnails from a pre-built 8-template Canva library at 75-90 minutes, batch-export and schedule via Buffer / Later / Metricool / Hypefury from 90-120 minutes. Two hours start to finish, 48 styled variants across 4 platforms ready to drip out over the next 14 days. The bonus folder includes 30-day-repurposing-calendar.md — a full publishing schedule template that maps each clip to a specific day and platform across a 4-week cycle, with re-cut rules for top performers and analytics-pass instructions.

Where this scales

The cadence the book proposes is 14 pieces per week across four platforms: 3 YouTube Shorts, 4 TikToks, 2 Reels, 5 LinkedIns. That works out to roughly 60 pieces over a 30-day cycle from a single weekly batch session plus one weekly long-form recording. A typical creator running this for 12-16 weeks sees the compounding kick in around week 12: the library has 100+ clips, search discovery starts compounding (a strongly-searchable clip adds 30-80 percent to its lifetime views from in-platform search plus external Google discovery, beyond the first-48-hour algorithmic push), the algorithm has enough data to know your patterns, and the long-to-short funnel starts producing real customers. Most creators quit at week 4-6 because the early returns look modest. The math runs differently after week 12 — the work is uphill until then, then it tilts.

The funnel math at scale: a creator running 100,000 monthly clip views across the four platforms generates roughly 1,500-3,000 new followers, 800-2,500 long-video clickthroughs, 80-300 email subscribers, and 2-15 new customers per month from the clip funnel alone. A creator who scales to 1M monthly clip views (entirely achievable after 9-18 months of consistent batching) produces 20-150 new customers per month — at which point repurposing has become the primary growth channel and paid ads become optional rather than required.

The weekly analytics pass is the discipline that turns hit-or-miss into pattern-recognition. Every Monday, 15 minutes, sort the previous week’s 14 pieces into three buckets: winners (top 20 percent by views — study the hook, the topic, the framing, re-cut next month), average (middle 60 percent — keep producing similar), and underperformers (bottom 20 percent — analyze hook and caption and topic, decide whether the source material is wrong or the framing failed). After 12-16 weeks of this, your hit rate (clips that significantly outperform baseline) climbs from roughly 1-in-20 at the start to 1-in-5. The library of winners grows. Re-cuts of winners become a meaningful share of monthly output, performing at 50-70 percent of the original — still above baseline for a new clip.

Included with the book

• Clip-scoring spreadsheet (CSV) — a 36-row template tracking source video, timestamps, hook and payoff scores, platform fit, and publish status. Drop in your candidates from the transcript pass and the prioritization sorts itself.
• 30-day repurposing calendar (markdown) — printable 4-week publishing schedule with day-by-day platform slot assignments, re-cut rules, and the weekly analytics pass template.
• Clip variant templates (markdown) — twenty Hook + Title + Caption templates, five per platform, with replacement variables for keyword, stat, outcome, and handle.

Get the full picture

Readers of this also chose

Questions readers ask

I have 200 hours of recordings but most of it is coaching calls with clients. Can I clip those?

Only if your consent forms cover derivative public content. Most coaching-call consent forms cover the original recording for client review but not public clips. The book walks through the permission audit in Chapter 2. For old recordings without explicit derivative-content consent, email the client and get a one-line yes in writing. Going forward, update your standard intake to include "and short-form clips for promotional and educational use across video platforms" — that single phrase removes 90 percent of the friction.

Do I need the full paid stack to start?

No. The book covers a free path using Whisper.cpp for transcription, CapCut Desktop for editing and captions, Canva free tier for thumbnails, and YouTube's auto-captions edited in YT Studio for accessibility tracks. The free path takes 30-50 percent longer per clip but is genuinely complete. The paid stack ($44-73/month for Submagic + Descript + Opus) saves 6-10 hours per month for creators publishing 4-8 clips per week. At any hourly rate above $5-10, the paid stack pays back.

What if my long videos are audio-only podcasts?

Audio-only podcasts clip but cost more time per clip. The trick is adding a visual layer — waveform animations, captioned key phrases, B-roll, or a static portrait of the host with animated captions. Submagic and Captions both support audio-to-vertical-video conversion. The trade-off is roughly 50-80 percent more time per clip versus video-source content. Many top podcast creators do this and it works. Your score-3-visual-interest videos can still produce strong clips; you just budget the extra time.

How long until the workflow pays back the setup time?

The setup investment is the library audit (90 minutes), the first batch session learning curve (90 minutes versus 60 minutes for subsequent sessions), and building the eight thumbnail templates in Canva (2 hours one-time). Roughly five hours of setup. The payback is in the per-clip time compression: instead of 60-90 minutes per clip you are at 10-12 minutes per clip in batch mode. At twelve clips per week the time savings cross the five-hour setup investment by the end of the second week.

What if I need a refund?

Checkout runs on Lemon Squeezy. The standard refund window applies. You keep the PDF either way.