When to use it
Use this after a webinar, podcast, customer interview, or recorded talk to get a week of content out of a single recording instead of letting a great asset die the day after the live event. It earns its keep when the recording actually contains sharp, non-obvious ideas worth repeating, a strong guest, a real customer story, a contrarian take. Run it in a large-context model so you can paste the full transcript, not a summary.
Do NOT feed it a rambling, low-insight recording and expect gold; if the source has no real ideas, the output will be confident filler. And do NOT publish the pull-quotes or stats without checking them against the recording, misquoting a guest, a customer, or your own exec is a fast, public way to lose trust. The model lightly tidies quotes for readability, which is fine, but it must never invent a quote or a number that isn't in the transcript.
The principle: repurposing means redistributing the original point of view across formats, not sanding it into generic 'thought leadership.' The most common failure is taking a spiky, specific talk and turning it into safe blog mush that says nothing. Preserve the edge, keep the best lines as actual quotes, and let each asset carry one sharp idea from the source rather than blurring all of them together.
The prompt
Prompt, paste into Claude, ChatGPT, Gemini
You are a content strategist who repurposes long-form recordings into multiple sharp assets without inventing anything. Every claim, quote, and stat must come from the transcript. If something isn't in the transcript, you leave it out. You preserve the speaker's actual point of view and edge, you never sand a spiky idea into generic thought-leadership.
TRANSCRIPT
{{TRANSCRIPT}}
CONTEXT
Topic / working title: {{TOPIC}}
Target audience: {{AUDIENCE}}
Brand voice: {{VOICE}}
Primary CTA / goal: {{CTA}}
TASK
Extract the best ideas and repurpose them into a set of assets, all grounded only in the transcript.
METHOD
1. First, extract the 3-5 most valuable, NON-OBVIOUS ideas in the transcript, the things worth repeating. Skip the generic ones.
2. Preserve the original POV and any contrarian edge; do not neutralize a strong take.
3. Build each asset around the source's real ideas, quotes, and moments. Lightly tidy quotes for readability and mark them (lightly tidied); never alter their meaning.
4. Do not add any fact, stat, customer, or claim not present in the transcript.
OUTPUT FORMAT
KEY IDEAS (3-5 bullets)
1. BLOG POST OUTLINE: SEO-friendly H1, a one-paragraph intro, 4-6 H2 sections each with 1-2 supporting bullets drawn from the transcript, and a closing CTA.
2. THREE LINKEDIN POSTS: each built on a DIFFERENT key idea, each with a scroll-stopping hook and a real quote/moment from the transcript.
3. ONE NEWSLETTER EMAIL: subject line + ~120-word body that teases the single best insight and drives the CTA.
4. FIVE PULL-QUOTES: short, punchy lines from the transcript, with speaker attribution, marked (lightly tidied) where edited.
5. THREE CLIP IDEAS: a timestamp or moment description + a one-line caption each. Flag the single strongest 30-second clip.
CONSTRAINTS
- Nothing outside the transcript. No invented facts, stats, customers, or quotes.
- No hype words (unlock, leverage, supercharge, seamless, game-changer, revolutionize).
- Keep the speaker's voice and POV intact; flag anything you tidied.
- If the transcript is thin on real ideas, say so honestly and produce fewer, stronger assets rather than padding.Run it from the terminal
$# paste a full transcript and repurpose it in one large-context call
$wc -w transcript.txt
8742 transcript.txt
$# concatenate the prompt template with the transcript, then pipe to Claude
$cat prompts/repurpose.md transcript.txt | llm -m claude-sonnet-4-6
KEY IDEAS
- CAC problems are usually conversion problems, not spend problems.
- Most teams track channel CAC but ignore stage-by-stage drop-off.
- One high-intent channel often beats five mediocre ones.
1. BLOG POST OUTLINE
H1: Your CAC Isn't a Spend Problem. It's a Conversion Problem.
H2: Why channel CAC hides the real issue (quote on 'optimizing the wrong number')
H2: The stage-by-stage drop-off audit
H2: Concentrate, don't diversify
2. LINKEDIN POSTS
- Post A: 'Cutting ad spend almost never fixes CAC. Here's what does.'
- Post B: 'You're measuring channel CAC. The leak is one layer down.'
- Post C: 'Diversifying channels can hide that none of them work.'
4. PULL-QUOTES
- 'You don't have a spend problem, you have a conversion problem.', [Speaker]
- 'Channel CAC is the number everyone watches and lies the most.', [Speaker] (lightly tidied)
5. CLIP IDEAS
- The 'channel CAC is a lie' moment. STRONGEST 30-SEC CLIP, lead with this.
$# everything grounded in the transcript; spot-check quotes before publishing
$
Variables to swap
| Variable | What to put there |
|---|---|
{{TRANSCRIPT}} | The full transcript of the webinar, call, or talk |
{{TOPIC}} | The topic or working title |
{{AUDIENCE}} | Who the content is for |
{{VOICE}} | Your brand voice (e.g. direct, practical, no-hype) |
{{CTA}} | The primary action you want readers to take |
Worked example
Input
Filled-in example
TRANSCRIPT: [full webinar transcript on cutting CAC in B2B SaaS, with a guest who argues spend cuts are the wrong reflex]
TOPIC: How to lower CAC without slashing growth
AUDIENCE: B2B SaaS marketing and growth leaders
VOICE: Practical, direct, a little contrarian
CTA: Subscribe to the newsletterOutput
What the model returns
KEY IDEAS
- CAC problems are usually conversion problems, not spend problems.
- Most teams track channel CAC but ignore stage-by-stage drop-off, where the real leak is.
- One high-intent channel often beats five mediocre ones; diversification can hide weakness.
1. BLOG POST OUTLINE
H1: Your CAC Isn't a Spend Problem. It's a Conversion Problem.
Intro: Most teams react to rising CAC by cutting budget. The talk argues the leak is almost always downstream of the ad, in the funnel, not the spend.
H2: Why channel CAC hides the real issue (quote on 'optimizing the wrong number')
H2: The stage-by-stage drop-off audit (the practical method from the talk)
H2: Concentrate, don't diversify (the contrarian take on five mediocre channels)
H2: What to fix first
CTA: Subscribe for the full teardown.
2. LINKEDIN POSTS
- Post A, hook: 'Cutting ad spend almost never fixes CAC. Here's what does.' Built on the conversion-vs-spend idea + the 'wrong number' quote.
- Post B, hook: 'You're measuring channel CAC. The leak is hiding one layer down.' Built on the drop-off audit.
- Post C, hook: 'Diversifying channels can be a way to hide that none of them work.' The contrarian concentration take.
3. NEWSLETTER EMAIL
Subject: your CAC problem is downstream
Body: ~120 words teasing the conversion insight and the drop-off audit, holding back the method, ending on Subscribe.
4. PULL-QUOTES
- 'You don't have a spend problem, you have a conversion problem.', [Speaker]
- 'Channel CAC is the number everyone watches and the one that lies the most.', [Speaker] (lightly tidied)
- ...3 more from the transcript
5. CLIP IDEAS
- The 'channel CAC is a lie' moment, caption: 'Stop optimizing the wrong number.' STRONGEST 30-SEC CLIP, lead with this for the video editor.
- The drop-off audit walkthrough, caption: 'The 10-minute audit that finds your real CAC leak.'
- The contrarian concentration take, caption: 'Five mediocre channels < one good one.'Tips to get more out of it
- Use a large-context model and paste the WHOLE transcript, summarizing first strips out the quotable, specific moments that make the assets worth publishing.
- Spot-check every pull-quote against the recording before publishing. 'Lightly tidied' is fine; changing what a guest or customer actually meant is a trust-destroying mistake that travels fast.
- Stagger the publishing schedule so one recording fuels two weeks of distribution, not one busy afternoon where everything competes with itself.
- Make it flag the single strongest 30-second clip so you know exactly where to point the video editor first, that one clip is usually the whole reason to repurpose.
- Feed the best LinkedIn post into the ad-copy prompt to test it as paid distribution; a line that performed organically is a strong paid hook candidate.
- If the transcript is genuinely thin, let the model tell you, producing three strong assets beats five hollow ones, and padding a weak recording just publishes filler under your name.
- Across models: Claude and Gemini handle the largest contexts well for full transcripts; if you hit a context limit, paste the transcript in sections and ask it to extract key ideas per section first, then assemble.