You have a real lesson worth sharing. But every draft reads like a humblebrag (“getting laid off was the best thing that ever happened to me”) or boilerplate motivation (“hard work pays off”). This page gives you a copy-ready AI prompt that turns a raw moment into a post built for the 2026 LinkedIn feed: a hook that survives the fold, a specific story, a non-obvious lesson, and a question people actually answer.
TL;DR
- Use AI for structure and cringe-cutting, never for your voice or the facts of the story.
- The first ~210 characters (desktop) or ~140 (mobile) are all most people see before “see more” — load the hook there, in one paragraph.
- LinkedIn’s 2026 ranking is driven by dwell time and substantive comments, not likes: posts that hold a reader 61+ seconds hit ~15.6% engagement vs. ~1.2% for a 0–3 second skim (AuthoredUp / dwell-time data, June 2026).
- Claude Sonnet 4.6 tends to produce the most natural prose; GPT-5.5 is fine if you feed it a strong voice sample and ban the AI tells (em dashes, “It’s important to note”, three-word paragraphs).
- Aim for 1,300–1,900 characters; that band is the current engagement sweet spot.
The prompt
Paste this into Claude or ChatGPT. Replace each [bracket] with your own material — the more specific your raw moment, the less generic the draft.
Draft a LinkedIn thinking post.
The raw moment (5-10 sentences: what happened, who was there, what
was said, how you felt — concrete, not summarized):
"""
[paste your raw moment]
"""
My audience: [who follows me, what they do]
My takeaway (draft): [your best guess at the lesson — or write
"propose 3 lesson framings and let me pick"]
Voice anchors (1-2 of my past posts that performed well):
"""
[paste 1-2 posts]
"""
Hard constraints:
- No humblebrag (never frame a setback as secret luck).
- No generic motivation ("hard work pays off", "trust the process").
- No corporate jargon ("synergies", "leverage", "circle back").
- No em dashes. Use periods or commas.
- No "Here's what I learned" / "Let that sink in" cliches.
- Plain words. No "delve", "tapestry", "in today's landscape".
Structure:
1. Hook: 1-2 lines, ONE paragraph (no line break), a specific
moment, present tense if possible. This must work as the only
thing a reader sees before "see more".
2. Story: 3-5 short paragraphs, concrete details and real numbers.
3. Lesson: 1-2 lines, non-obvious. Not the takeaway the reader
predicts from the hook.
4. Question: open and honest, something I'd genuinely want answered,
not rhetorical.
Length: 1,300-1,900 characters. Mobile-readable line breaks, but
keep the first paragraph unbroken so the hook isn't cut off.Why these constraints, not generic ones
The prompt above is tuned to how LinkedIn actually ranks and renders posts in 2026, not to vibes.
| Rule in the prompt | Why it’s there (June 2026) |
|---|---|
| Hook in one unbroken paragraph | A double line break almost always ends the preview early; only ~210 chars (desktop) / ~140 (mobile) show before “see more”. |
| Ban em dashes + AI cliches | Readers now openly hunt for AI tells; em dashes and “It’s important to note” are the loudest flags and tank credibility. |
| Open question at the end | Comments weigh ~15x more than likes, and LinkedIn only counts comments of roughly 5+ words, so you need a question worth a paragraph. |
| 1,300-1,900 characters | This range shows the strongest engagement; shorter posts skim, much longer ones lose dwell time. |
| Concrete story over “a realization” | Dwell time is the primary quality signal; specific scenes hold readers, summaries get scrolled past. |
| No external links in the post itself | LinkedIn suppresses reach on posts with outbound links; put any link in the first comment instead. |
Which model to use
Both major models write competent drafts. The difference is how much editing you do afterward.
| Model (June 2026) | Best for | Watch out for |
|---|---|---|
| Claude Sonnet 4.6 (Free / Pro $20) | Most natural default prose, fewest AI tells | Can still over-polish; tell it to keep your rough edges |
| GPT-5.5 Thinking (Plus $20) | Strong structure, good at 3-framing brainstorms | Defaults to em dashes and tidy “lessons” — keep the ban list |
| Gemini 3.1 Pro (Google AI Pro $19.99) | Long voice samples (1M context) | Tends toward formal tone; feed casual anchors |
For a thinking post, run the brainstorm in Claude or GPT-5.5, pick a lesson framing, then ask for one tight draft. Don’t accept the first output as final.
How to check the draft before you post
- Read it aloud. If a sentence makes you wince, cut it. This catches more cringe than any rule.
- Test the fold. Copy just the first paragraph into a LinkedIn character counter — it should land its hook inside ~210 characters with no line break.
- Scan for tells. Search the draft for em dashes, “It’s important”, “Here’s the thing”, and any three-word dramatic paragraph. Delete them.
- Pressure-test the lesson. Could a reader predict it from line one? If yes, the lesson is too clean — real lessons have edges and exceptions.
- Check the question. Would you personally write a 5-sentence answer to it? If not, it won’t earn the comments that drive reach.
Common mistakes
- Humblebrag in a vulnerability costume. “I almost quit” that resolves into “and that’s how I 10x’d revenue” reads as a brag. Keep the unresolved part.
- Lessons that sound like fortune cookies. If it fits on a mug, push the model to find the uncomfortable version.
- No actual scene. “Last week I had a realization” is not a story. “We were 12 minutes from launch and the API key was wrong” is — give the model that level of detail.
- Posting daily for volume. Once a week of substance beats five filler posts; dwell time rewards depth, and the algorithm scores each post hard in its first hour.
- Pasting the AI draft verbatim. The voice and the facts have to be yours. AI handles the shape; you supply the experience nobody else has.
FAQ
Should I disclose that AI helped? Not required. The substance and voice are yours; the model only proposed structure. Disclose if your audience or employer expects it.
Where do I put a link? In the first comment, not the post body — outbound links in the post itself reduce reach. Drop the link below and reference it (“link in comments”).
How long should the post be? Target 1,300–1,900 characters. The hard ceiling is 3,000, but past ~2,000 dwell time and completion drop.
Will an AI detector flag it? Possibly, and that’s the wrong target. Detectors aren’t perfect, and readers’ eyes are the real filter. Cut the tells, add your specifics, and it will read as yours.
When should I post? Tuesday–Thursday, around 8–9 AM or 2–3 PM in your audience’s timezone, then reply to early comments — the first hour of engagement decides distribution.