Content Repurposing Prompts: One Piece, Twelve Outputs

TL;DR: A content repurposing workflow turns one long article into platform-native outputs: thread, LinkedIn post, newsletter, video script, carousel, FAQ, quote cards, each rewritten with its own opening argument rather than shortened from the source. Twelve prompts below map to specific source sections, and the last one covers saving them so next month's article takes minutes, not another afternoon.

What Is a Content Repurposing Workflow (and Why Does the Naive Version Read as Spam)?

A content repurposing workflow takes one piece of source content (usually a long article, a video transcript, or a research piece) and produces a set of shorter, platform-specific pieces from it, instead of writing each platform's version from scratch. The idea itself isn't new. What's usually missing is the one distinction that separates a workflow from a repetition machine: repurposing has to change the argument's shape, not just its length.

Naive repurposing takes the same paragraph and trims it. The thread gets the article's opening sentence split across three posts. The LinkedIn post gets the same sentence with two extra lines of context. The newsletter gets the same sentence again, maybe reordered. Anyone who follows the same person in more than one place sees the identical claim, phrased almost identically, three times in one week. That's not a system. That's spam with better line breaks.

Say the source article opens with this line (illustrative for this example; swap in your own measured numbers if you've actually got them):

"Most teams adopted async standups to save meeting time. But when you actually time it, async takes longer than the meeting it replaced. The writing takes longer than the talking did, and most of what gets written gets skimmed, not read."

Bad (same claim, same shape, twice):

Thread, post 1: "Most teams adopted async standups to save meeting time. But when you actually time it, async takes longer than the meeting it replaced."

LinkedIn, opening line: "Most teams adopted async standups to save meeting time. But when you actually time it, async takes longer than the meeting it replaced. The writing takes longer than the talking did."

Good (same source, two different argument shapes):

Thread, post 1: "Async standups take 12 minutes to write and 40 seconds to read. That's not a time save. It's a time transfer, from the person who used to talk to the person who now has to type."

LinkedIn, opening line: "I killed our daily standup to 'save time.' Then I actually timed the async version that replaced it. I owe my team an apology."

The thread version leads with the paradox as a stat, built for a scroller who wants the punchline before the second sentence. The LinkedIn version leads with a personal admission, built for a reader who responds to a confession and a discussion, not a stat dropped cold. Same underlying claim, same source article. Two different entry points, because different platforms are different rooms with different expectations for how you walk in.

What Source Article Does This Workflow Assume?

The twelve prompts below assume a single roughly 2,000-word article with a clear argument, not a listicle or a roundup. Repurposing needs a thesis to redirect, and a listicle doesn't have one to redirect. Assume it breaks into five sections doing five different jobs:

1. The hook: the counterintuitive claim, stated once, in one or two sentences.
2. The evidence: the data, math, or example backing the claim.
3. The mechanism: why the problem happens, not just that it happens.
4. The exception: where the claim doesn't hold, stated honestly.
5. The recommendation: what to do instead, and how to start.

Every prompt below names which of these five sections it needs. That's the actual difference between "repurpose this article", which is vague and produces section 1 restated twelve times, and a prompt that tells the model exactly which section to pull from and what shape to force it into.

How Do You Map One Article to Twelve Outputs?

Before opening a single AI tool, decide what each output needs from the source and how long it can run. This is the map. Build every prompt below around it rather than improvising per output.

Two things fall out of this table immediately. The shortest outputs (hooks, subject lines) draw only from the hook section, so don't force the evidence in; there's no room and it reads as clutter. And the longest output, the video script, is the only one that needs all five sections in the original order. Everything else here is deliberately partial.

How Do You Turn the Article Into an X Thread, a LinkedIn Post, and a Newsletter Section?

All three are read as text on a feed, but they differ in trust level and skim speed, which is why the same source line can't lead all three.

X thread. Draws from the hook and the evidence; needs the paradox as post one, not the article's actual opening sentence.

Turn this article into a 7-post X thread.

ARTICLE: [paste full article]

Post 1: the single most counterintuitive claim from the piece, phrased as a
stat or a paradox — not the article's actual opening sentence. Under 200
characters (the free-tier limit is 280; leave room to breathe).
Posts 2-3: the evidence that backs post 1 — one data point or example each.
Posts 4-5: the mechanism — why this happens, in plain language.
Post 6: the one honest exception, stated straight, no hedging.
Post 7: what to do instead, plus a question that invites a reply.
Do not reuse a full sentence from the article verbatim — rewrite each claim
for a scroller who has 2 seconds and no surrounding context.

What carries over: the underlying facts and the mechanism. What must change: the opening line and sentence length. Each post has to work read alone, mid-scroll, by someone who never saw the article.

LinkedIn post. Draws from the hook and the mechanism; needs a personal framing, not a stat.

Turn this article into a LinkedIn post.

ARTICLE: [paste full article]

Opening line (this is what shows before "see more" on mobile — keep it
under 140 characters): a personal admission or a specific moment, not a
general claim.
Body (150-220 words): the mechanism from the piece, told through what I
actually did or noticed — first person, not "studies show."
Close with one question that invites a comment, not a link.
Do not open with the same sentence or stat used in the X thread version —
this reader hasn't seen that one and shouldn't feel like they're getting a
recycled post if they follow both accounts.

What carries over: the mechanism. What must change: point of view (first person, not third), and the fact that the hook has to survive LinkedIn's mobile truncation, which lands around 140 characters before "see more."

Newsletter section. Draws from all five sections but heavily compressed, and assumes trust already exists, so it skips the hook entirely.

Turn this article into a 150-word section for my existing newsletter.

ARTICLE: [paste full article]

No hook, no re-introduction of the topic — my subscribers already trust me,
so open with the takeaway in sentence one.
One line only a subscriber would get: reference something I've mentioned
before, or a detail too specific for a cold audience.
End with one link-out line: "[link text] →" pointing to the full piece.
Keep it to one paragraph. No bullet points.

What carries over: the takeaway and the mechanism. What must change: remove the hook entirely. A subscriber already opted in, so re-hooking them reads as filler they'll skip.

How Do You Adapt the Article Into a YouTube Script Outline and Three Short-Form Video Hooks?

Video is the one medium here where a headline doesn't work as a hook. A hook has to be something a viewer hears or sees in the first two to three seconds, not something they read.

YouTube script outline. Needs all five sections, restructured as narrative rather than argument.

Turn this article into a YouTube script outline for an 8-10 minute video.

ARTICLE: [paste full article]

Cold open (0:00-0:20): a scene or moment, not a stated claim — show the
problem happening before naming it.
Section 1 (the claim): state the counterintuitive point directly, once.
Section 2 (the proof): walk through the evidence like it's being discovered
on camera, not recited from notes.
Section 3 (the mechanism): the "why," in plain spoken language —
contractions are fine, this is spoken, not written.
Section 4 (the exception): say where the claim is wrong, on camera — this
is the moment that builds trust with a video audience specifically.
Section 5 (the fix + CTA): what to do differently, then one ask.
Mark each section with a timestamp and a one-line visual/B-roll note.

For the platform-specific scripting conventions this outline feeds into — pacing, retention hooks, thumbnail-title pairing — see our YouTube prompts guide rather than rebuilding that layer here.

What carries over: the five-part argument, reordered as narrative. What must change: everything becomes spoken and visual. No bullet point from the article survives into a script meant to be read aloud.

Three short-form hooks. Same source claim, three different opening conventions, because TikTok, Reels, and Shorts audiences arrive with different expectations even watching the identical clip.

Write three 3-second spoken hooks for the same short-form clip, all built
from this article's core claim.

ARTICLE: [paste full article]

TikTok hook: pure curiosity gap, under 8 words, no context given — the kind
of line that only makes sense once the viewer keeps watching.
Instagram Reels hook: pairs with a visible action or prop in frame — under
10 words, written to be said while doing something on camera.
YouTube Shorts hook: allows a half-second longer runway — under 15 words,
can set up a mini scenario before the twist.
Do not write the same hook three times with synonyms swapped — each must
use a different rhetorical move (question, contradiction, confession).

What carries over: the core claim only. What must change: the rhetorical device itself. A question, a stated contradiction, and a confession are three different tools, not three phrasings of one tool.

How Do You Turn the Article Into an Instagram Carousel and an Email Subject Line Pair?

Instagram carousel. Needs the full argument restructured as one idea per slide, claim-first.

Turn this article into a 7-slide Instagram carousel.

ARTICLE: [paste full article]

Slide 1: the boldest single-sentence claim from the piece. Under 12 words,
big enough to read paused mid-scroll.
Slides 2-3: the evidence, one point per slide, under 15 words each.
Slide 4: the mechanism, one sentence.
Slide 5: the honest exception.
Slide 6: the recommendation.
Slide 7: one CTA line + "swipe back if you missed it."
Each slide is a standalone sentence — no slide should require the previous
one to make sense if someone opens the carousel mid-way through.

What carries over: the five-part structure, one idea per slide. What must change: every slide has to stand alone visually; carousels get viewed out of order more than any other format here.

Email subject line pair. Draws only from the hook; needs two genuinely different angles for an A/B test, not two synonyms of the same angle.

Write two email subject lines for a newsletter linking to this article,
built for an A/B test — not two versions of the same angle.

ARTICLE: [paste full article]

Subject A: curiosity-driven, withholds the answer, under 40 characters
(most inbox previews truncate well before 45 on mobile).
Subject B: benefit-driven, states the payoff directly, under 40 characters.
The two must use different rhetorical strategies — don't give me two
curiosity subject lines with different words.

What carries over: nothing but the underlying topic. Subject lines don't inherit sentences from the article, only the hook. What must change: length is the hard constraint; the whole idea has to survive in well under 10 words.

How Do You Extract a FAQ Block, a Quote Card Set, and a Slide Outline From the Same Article?

FAQ block. The one output here that doesn't need an argument shape at all; it needs direct, standalone answers.

Turn this article into a 6-question FAQ block.

ARTICLE: [paste full article]

Write 6 questions a skeptical reader would actually ask after skimming the
headline — not questions the article already answers in the same order.
Each answer: 40-90 words, standalone — answerable without having read the
rest of the FAQ or the article itself.
No question should restate another in different words.

What carries over: the underlying facts, redistributed. What must change: everything. A FAQ has no narrative and no hook. It's the one format here where "the same paragraph rephrased" is actually fine, because each answer is supposed to be self-contained by design.

Quote card set. Pulls the sharpest individual lines and doesn't paraphrase them down to nothing.

Pull 4 quote-card lines from this article.

ARTICLE: [paste full article]

Each line: under 20 words, a complete thought with no pronoun that needs the
surrounding paragraph to resolve ("it," "this," "that" referring to
something outside the line itself).
Prefer the sharpest existing sentence over inventing a paraphrase — quote
cards work because they sound like something a person actually said, not
like ad copy.
One of the four should be the exception, not the core claim restated a
fourth time.

What carries over: nearly verbatim wording. This is the one output that should not be heavily rewritten, unlike every stage above it. What must change: selection, not phrasing. Pick lines that survive total removal of context.

Slide outline. For a live talk or webinar deck; sequential logic for a synchronous audience, unlike every output above it.

Turn this article into a 10-slide deck outline for a live talk.

ARTICLE: [paste full article]

Slide 1: title + the claim, posed as a question to the room.
Slides 2-3: the problem, made concrete with the evidence.
Slides 4-5: the mechanism, one idea per slide.
Slide 6: the exception — say it before someone in the room brings it up.
Slides 7-8: the recommendation, broken into two concrete steps.
Slide 9: what happens if nothing changes.
Slide 10: one ask.
For each slide: a headline under 8 words, plus one line of speaker notes.

What carries over: the five-part logic, in its original order: the one output here where order matters as much as it did in the source. What must change: pacing for a room that can't skip ahead. A live audience can't scroll past the slow slide, so nothing here can drag.

How Do You Make This Workflow Reusable Next Month?

Building twelve prompts once is worth an afternoon. Retyping them every time a new article publishes is not. That's the point where a repurposing workflow either becomes a system or quietly stops happening after the second month.

The mechanical fix is saving each of the twelve prompts above as a template with [ARTICLE] as the one variable that changes, instead of as a note in a doc you have to find and hand-edit every time. Prompt Architects' Prompt Library does exactly that: each prompt gets saved once, and a Global Variable holds the placeholder. Next month, paste the new article into the variable and run all twelve saved prompts against it, instead of rebuilding the thread prompt, the carousel prompt, and the other ten from memory.

That's not an argument for this specific tool over a working notes doc. A doc with all twelve prompts pasted in, and a placeholder you swap the new article into by hand, does the same job if it's genuinely working for you. What a saved library changes is not doing the job better in principle, just not re-finding and re-editing twelve prompts every time, with them available inside ChatGPT, Claude, or Gemini directly instead of tabbing back to a doc mid-session.

For the broader pattern of building one prompt and running it against new inputs repeatedly, instead of writing a fresh prompt every time, see how a saved prompt template works in our code review prompt generator, and how a growing set of saved prompts stays usable in organizing 1,000+ prompts. If a repurposed draft comes back flat on the first pass, running it through a free prompt enhancer is a faster fix than rewriting the prompt from scratch. And if different stages get routed to different models (a creative hook to one, a precise FAQ answer to another), the sampling differences between providers are covered in the LLM parameter cheat sheet.