The bottleneck in content marketing isn't ideas — it's production. You write a 2,000-word blog post and it gets published once. The same information could have driven 30 pieces of content across six platforms that week. Most of it doesn't require new thinking; it requires reformatting.
AI is genuinely good at this. Repurposing is a mechanical transformation — take the core ideas and adapt them to a new format's rules. That's a solvable prompting problem. The tricky part is maintaining your voice across formats, and not producing content that sounds like it was written by someone who read a summary of your post.
Here's the workflow and every prompt I use.
The repurposing mindset
The mistake most people make: they paste their whole blog post into ChatGPT and say "make this into social posts." The output is terrible because the AI is averaging across the entire piece instead of extracting specific ideas.
Good repurposing is selective. A 1,500-word blog post might have 3 tweet-worthy insights, 1 LinkedIn story, and 1 carousel concept — not 10 of each. You need to identify the best fragments, then prompt specifically for each one.
The other thing to get right: format-fit. An Instagram caption is not a shortened blog paragraph. A Twitter thread is not a listicle with line breaks. Each format has its own grammar. Your prompts need to specify that grammar explicitly.
What converts to what
Not all source content repurposes equally well. Here's what works:
Blog post → Twitter/X threads, LinkedIn posts, email newsletter sections, Instagram carousels, YouTube scripts (with expansion prompts), FAQ pages, podcast talking points
Podcast episode → Blog post, show notes, email newsletter, key quote graphics, YouTube short clips, LinkedIn insights
Video/webinar → Blog post, carousel scripts, email series, transcript-based SEO content, TikTok/Reels clips with text
Talk or presentation → Long-form LinkedIn article, blog post, email sequence, Twitter thread
The common pattern: video and audio content goes up the production chain (toward text and structured formats), and written content goes down (toward shorter, more casual social formats).
The master repurposing prompt framework
Every repurposing prompt shares the same structure. Fill in the variables:
SOURCE CONTENT TYPE: [blog post / podcast transcript / video transcript / email]
SOURCE CONTENT: [paste or describe]
OUTPUT FORMAT: [Twitter thread / LinkedIn post / Instagram carousel / email newsletter / etc.]
TARGET AUDIENCE: [who should read this]
TONE: [match my original / more casual / more formal / punchier]
VOICE FINGERPRINT: [3-5 sentences describing how you write — see LinkedIn post for how to generate this]
Format rules for [OUTPUT FORMAT]:
[Paste the specific format rules from the relevant section below]
Additional constraints: [any specifics — length, avoid certain phrases, include a CTA for X]
Prompts by output format
Twitter/X thread
Turn the key ideas from this [SOURCE CONTENT TYPE] into a Twitter/X thread.
Source: [PASTE CONTENT OR SUMMARY]
Rules:
- Tweet 1: the hook — the single most counterintuitive or valuable insight from the whole piece. Under 240 characters.
- Tweets 2-[N]: one idea per tweet. State the point, then one supporting sentence. Under 240 characters each.
- Final tweet: a brief takeaway + one of: question for replies, link to original, or CTA
- Maximum 10 tweets (including hook and closer)
- Do NOT number the tweets (no "1/10", "2/10")
- No thread intros like "A thread:" or "🧵"
Extract only the ideas strong enough to stand alone — don't try to cover everything.
Example output hook (from a blog post about pricing):
"Raising your prices is the easiest marketing decision you'll make this year. Here's why lower prices make your marketing harder, not easier."
LinkedIn post (story format)
Extract one specific story, example, or moment from this content and turn it into a LinkedIn story post.
Source: [PASTE CONTENT]
Rules:
- Start in the middle of the action — not at the beginning
- Keep to 150-200 words
- End with one insight, not a list of lessons
- Use single-line breaks throughout
- Don't include hashtags
Voice: [PASTE YOUR VOICE FINGERPRINT]
LinkedIn post (insight format)
Find the single most actionable or counterintuitive idea in this content:
[PASTE CONTENT]
Write a LinkedIn insight post around that idea.
- Lead with the conclusion
- First line should be strong enough to stop a scroll
- Under 100 words
- Specific — no vague wisdom
- Don't hedge
Voice: [PASTE YOUR VOICE FINGERPRINT]
Instagram carousel
Turn this content into an Instagram carousel script:
[PASTE CONTENT]
Slide structure:
- Slide 1 (hook): One bold statement or question. Under 8 words. Creates curiosity without giving away the content.
- Slides 2-7: One idea per slide. Format: short headline + 1-2 supporting sentences. Each slide should be able to stand alone.
- Final slide: Key takeaway + CTA (save this / follow for more / link in bio)
Rules:
- Each slide: under 30 words total
- Build from less important to most important — save the best insight for slide 6
- No corporate language, no generic tips
Email newsletter section
Repurpose this content into a newsletter section for [AUDIENCE]:
[PASTE CONTENT]
Format:
- Section header (5 words max)
- Opening paragraph: 2-3 sentences that make the reader care about this topic right now
- 3-4 bullet points or a short numbered list: the most actionable takeaways
- Closing sentence: forward-looking or with a CTA
Length: 150-250 words
Tone: like you're writing to a smart colleague, not broadcasting to an audience
Don't restate everything — pick the 3 most useful ideas and go deep on those
YouTube script from blog post
Expand this blog post into a YouTube video script:
[PASTE BLOG POST]
Target video length: [DURATION] minutes
The script needs:
- A hook (0-30s) that doesn't start with "In this video" — open with the core problem or a surprising claim
- Spoken-word language throughout (no complex sentences designed to be read, not heard)
- A pattern interrupt every 90 seconds (mark these with [PATTERN INTERRUPT] in the script)
- Re-hooks before each new section ("Here's where it gets interesting" style — but less cliche)
- A CTA at the end (subscribe + tease next video)
For each section of the blog post, expand with: a personal anecdote, a concrete example, or a demonstration that doesn't exist in the written version.
Blog post from podcast transcript
I have this podcast transcript (or rough notes): [PASTE TRANSCRIPT/NOTES]
Write a blog post based on this material.
Rules:
- Don't transcribe the conversation — extract the ideas and restructure them as a well-organized article
- Lead with the most interesting insight, not with "In this episode"
- Use headers to organize the main points
- Where the transcript has a good quote or turn of phrase, keep it (in quotation marks if from a guest)
- Target length: [WORD COUNT]
- Add a brief intro that gives context for a reader who hasn't heard the episode
Tone: [match my blog voice — paste fingerprint]
TikTok/Reels script from long-form content
Extract the single most compelling moment, insight, or argument from this content and write a 30-second TikTok/Reels script:
[PASTE CONTENT]
Rules:
- Hook (first 2 seconds): one sentence, under 10 words, no setup
- Body (3-25 seconds): the core idea with one specific example or number
- CTA (last 5 seconds): one action, conversational tone
- Write for spoken delivery — short sentences, natural pauses
- Don't try to cover everything — pick ONE idea and make it land
Key quotes for social graphics
Extract 8 pull quotes from this content that would work as standalone social graphics:
[PASTE CONTENT]
Each quote should:
- Make sense without context (no "as I mentioned above")
- Be under 25 words
- Be specific — include a number or concrete detail where possible
- Sound like something a person said, not wrote
- Be genuinely interesting, not just accurate
Format: numbered list, quote only (no explanation needed)
Podcast show notes from transcript
Write show notes for this podcast episode based on the transcript:
[PASTE TRANSCRIPT]
Include:
- Episode summary (3-4 sentences, written for someone who hasn't listened)
- Key topics covered (bullet list, 5-7 items)
- 3 key takeaways (specific and actionable)
- Timestamps for the main topic changes (estimate based on the transcript flow)
- Guest bio (if applicable): [PASTE BIO OR DESCRIBE GUEST]
Tone: informative, not promotional — the show notes are for listeners deciding whether to tune in, not a marketing piece
FAQ from blog post
Read this blog post and generate an FAQ section:
[PASTE BLOG POST]
Generate 8 questions a reader might have after reading this post.
Include:
- Questions the post answers but doesn't explicitly state as questions
- Questions the post raises but doesn't fully answer (answer them in the FAQ)
- At least 2 "but what about..." questions that address objections
Format:
**Q: [Question]**
A: [Answer — 2-4 sentences, specific]
Maintaining voice across formats: the fingerprint prompt
The biggest problem with AI repurposing is that each format ends up sounding slightly different. The Instagram carousel sounds more casual than the LinkedIn post. The newsletter section sounds more formal than the tweet thread. Across a week of repurposed content, you sound like five different people.
Fix this with a voice fingerprint prompt, run once:
Here are 5 examples of my writing across different formats:
[PASTE EXAMPLES — mix of tweets, posts, emails, whatever you have]
Analyze my writing and produce a "voice fingerprint" — a description of how I write that I can paste at the start of any content prompt to ensure consistency.
The fingerprint should describe:
- My typical sentence length and variety
- Words or phrases I use (and ones I avoid)
- How I handle humor (if I do)
- My relationship to the reader (peer / teacher / collaborator / etc.)
- 3 example sentences in my voice that could serve as style anchors
Keep the fingerprint under 150 words — it needs to be paste-able without overwhelming the prompt.
Paste this fingerprint at the top of every repurposing prompt. It doesn't guarantee perfect consistency, but it reduces the drift significantly.
The complete repurposing workflow
Here's how this works end-to-end for a single piece of content:
Step 1 — Extract the ideas (run once):
Read this [CONTENT TYPE] and list:
- The 5 most interesting or counterintuitive ideas
- The 1-2 best stories or examples
- The single most actionable takeaway
- The most quotable sentence
[PASTE CONTENT]
Step 2 — Run format-specific prompts using the lists above, referencing the extracted ideas rather than re-pasting the full content each time.
Step 3 — Quality-check checklist:
- Does each piece stand alone (no references to "the original post")?
- Does each piece match the format's grammar (tweet length, carousel slide word count, etc.)?
- Does the voice feel consistent across pieces?
- Is at least one specific detail, number, or example in each piece?
Step 4 — Schedule and track: Which formats drove the most engagement back to the original piece? Double down on those. Stop producing formats that consistently underperform.
Setting this up with automation
For teams or high-volume creators, this workflow can be automated with n8n or Zapier:
- Trigger: new blog post published (RSS feed)
- Step 1 prompt runs automatically via Claude or OpenAI API — extracts ideas
- Format-specific prompts run in parallel — generate drafts for each output
- Drafts go into a review queue (Notion, Airtable, or a Google Sheet)
- You review, edit for voice, and approve for posting
The automation handles the generation. You handle the editorial judgment — which ideas are worth amplifying, which drafts need work, which channels are worth maintaining. For a deeper look at how to build multi-step AI pipelines like this, the post on AI research workflows covers the architecture.
The leverage here is real. A single good blog post, properly repurposed, can fuel two weeks of content across platforms. The prompts do the mechanical work. What they can't do is decide which ideas are worth your audience's time — that's still your call. For copy-paste templates on marketing content specifically, the marketing prompt library has ready-to-use versions of the most common formats. For more on prompting AI to write marketing content that actually converts, prompting for marketing covers the underlying strategy.
