The Content Repurposing Engine: A Teardown of the n8n Workflow That Multiplies One Post Into Five

This is part of my build-in-public series on running LLM workflows in n8n in production. Everything here is from a real template I sell and support — the mistakes are mine, the fixes are tested.

The pipeline in 30 seconds

One source post in → a distillation node that extracts the core idea, the proof, and the takeaway → five platform nodes that rewrite for X (thread), LinkedIn (post), a newsletter section, a short-form video script, and a community post → a portability check that catches what each platform can't render → a staggered 7-day calendar out. The point was never "post everywhere." It was: one idea, five native shapes, zero manual rewriting. The failures were all in what "native" turned out to mean.

Failure #1: the copy-paste costume

The first version technically produced five outputs. In practice it produced one text in five costumes — the same sentences, the same structure, the same voice, just line-broken differently. The X thread read like a LinkedIn essay chopped into pieces. The video script read like a blog post. Nobody would have been fooled for a second, and more importantly, nothing worked: each platform rewards native behavior and punishes imports.

The fix: per-platform voice contracts as hard constraints, not suggestions. The X node gets ≤280 chars per tweet, hook in the first 12 words, one idea per tweet, thread arc: hook → proof → twist → takeaway. The LinkedIn node gets hook line + personal story + 3-line CTA, no hashtags in the first line. The video script node gets spoken register, ~150 words per minute, visual beats in brackets. The model doesn't get to choose the shape — the shape is the node. Voice is a parameter, same as in the inbox-triage post: parameterized once, respected everywhere.

Failure #2: the same-day pileup

Even with good platform voices, the workflow shipped all five pieces at once. My audience saw the same idea on X in the morning, on LinkedIn at lunch, and in the newsletter that evening. Engagement on the third exposure wasn't additive — it was cannibalistic. Each piece stole attention from the others, and the comments told me: "saw this already."

The fix was to emit a calendar, not a content dump. The last node now produces a staggered 7-day schedule: day 1 the X thread (fast, discoverable), day 3 the LinkedIn post (leads with a different angle of the same idea), day 5 the newsletter section (the full reasoning, for the audience that wants depth), day 7 the video script (for the audience that never reads). Each piece opens with a different angle, so even someone who sees all five gets five doors into the same house, not the same room five times. Reach went from "one spike, four echoes" to a sustained week.

Failure #3: the broken transplant

This was the quiet one. Repurposed pieces kept carrying references that only made sense in the source: "as I showed in the table above" (there was no table on X), a link to a dev.to section anchor (dead on LinkedIn), code blocks (rendered as garbage in the newsletter preview), "reply to this tweet" (meaningless in a video script). The content transferred; it didn't transplant.

The fix: a portability check before anything ships. A validation node walks each output looking for the platform-incompatible: relative references ("above", "below", "in the thread"), links the platform can't render, formatting the platform mangles, and length overflows. Anything flagged goes back to the platform node with a one-line instruction: rewrite this reference self-contained. The rule the node enforces: every piece must make sense to someone who never saw the source. That's the real test of repurposing — not whether you can post it five places, but whether each place stands alone.

The cost math

One source post (~2k tokens) into five platform pieces (~1.5k output tokens each with prompts): roughly $0.05–$0.10 for the full set on GPT-4o-mini. The distillation node is the cheapest and the most important — if it extracts the wrong core idea, all five pieces inherit the mistake. I spend the expensive tokens there: a longer prompt with two worked examples of good vs. bad distillation. Five dollars a month of tokens for a week's worth of native content per post.

What I kept manual

Three things stay human: picking which posts deserve repurposing (not every post earned it — only the ones that already performed, because repurposing multiplies whatever you feed it, including mediocrity), the first line of each piece (the hook decides everything on every platform and takes me 30 seconds to sanity-check), and the actual posting (timing and replies are still mine). The workflow turns one good idea into a week's campaign in about 10 minutes. Deciding which ideas are worth a week is the job; the workflow just does the multiplying.

---

I'm building AI Automation Lab — production-ready n8n workflow templates. The AI Content Repurposing Engine is the full workflow from this article: the distillation node, the five per-platform voice contracts, the portability check, and the 7-day staggered calendar. My open-source patterns live at github.com/zhp910318/n8n-ai-automation-lab.