- Why most people's ChatGPT LinkedIn prompts don't work
- The prompt structure that changes the output
- Ten prompts I keep in my notes app
- A real example from my own feed
- A step-by-step routine that takes fifteen minutes
- Prompts for the formats people forget
- What I'd tell you if we were on a call about this
- Frequently asked questions
- Sources worth reading
The short version: a good ChatGPT prompt for LinkedIn gives the model your story, your data point, and your opinion, and asks for a draft you will rewrite, not a finished post you will paste. The prompts that fail are the ones that ask ChatGPT to "write a LinkedIn post about X" with nothing else, because that's when you get the same five hooks everyone else on your feed is already using this month.
Related reading: how long should linkedin profile summary be.
Why most people's ChatGPT LinkedIn prompts don't work
I've been posting on LinkedIn since before it had a feed algorithm worth mentioning, and I rebuilt a chunk of my business on that platform after five rough years of losing clients, losing speaking gigs, and losing the plan I'd had. So when I say I've tested a lot of prompts, I mean hundreds, across three accounts, over about eighteen months.
Here's what I noticed straight away. If you type "write me a LinkedIn post about leadership" into ChatGPT, you get a post that starts with a bold one-line hook, has three short punchy paragraphs, ends with a question, and reads like it was written by someone who has never led anyone. It's not a bad post. It's a nothing post. And LinkedIn is now so full of that exact nothing post, from the same three prompt templates people copy off YouTube thumbnails, that readers scroll past it in under a second.
The fix isn't a better single prompt. It's a different way of feeding ChatGPT your own material before you ask it for anything.
The prompt structure that changes the output
I use a five-part input every time, whether I'm writing about AI adoption, client work, or my own business mistakes. You can copy this structure straight into ChatGPT:
- 1. The specific moment. Not "I learned a lesson about pricing." Instead: "In March I quoted a client £4,500 for an AI workshop and they said yes in eleven minutes, which told me I'd underpriced it for two years."
- 2. The number or fact. A stat, a date, a price, a result. ChatGPT cannot invent your lived detail, so give it one.
- 3. Your actual opinion. Something you'd say out loud to a friend, including the bit that might annoy someone.
- 4. Who you're talking to. "Small business owners under 10 staff who are scared of AI, not marketing directors."
- 5. What you want them to do or feel. Not "engage," something real: "feel less stupid for not knowing this already."
Then the actual prompt is simple: "Using the details above, draft a LinkedIn post in my voice. Short paragraphs, no hashtags, no 'in today's fast-paced world,' end with a real question, not a generic one." That last line matters more than people think, because ChatGPT's default closing question is almost always "What do you think?" and nobody answers that.
If you want a deeper breakdown of exactly how to stop the output sounding synthetic, I wrote a full guide on ChatGPT prompts for LinkedIn posts that do not sound like AI, and a separate one with prompt variations by post type in LinkedIn post prompts for ChatGPT that don't read like everyone else's feed.
Ten prompts I keep in my notes app
These are copy-paste ready. Swap in your own detail before you use them.
- "Here's a mistake I made in my business: [detail]. Write a LinkedIn post that opens with the mistake, not a lesson, and gets to the lesson in the last two lines only."
- "I disagree with [common industry opinion]. Here's why: [your reasoning]. Write a post that states the disagreement in the first sentence, no throat-clearing."
- "Turn this client result into a post: [result, with numbers]. Do not use the word 'excited.' Do not start with 'I'm thrilled to share.'"
- "Write a post structured as five short numbered lessons from [event], each lesson one sentence, no explanation after each one."
- "I'm responding to a trend where people are saying [trend]. Write a contrarian post that respects the reader's intelligence and doesn't strawman the other side."
- "Take this paragraph from my own writing: [paste your own draft]. Tighten it, keep my exact phrases where they're strong, cut anything that sounds like a LinkedIn influencer wrote it."
- "Write a post that is just a list of things I used to believe about [topic] and don't anymore, one line each, no summary at the end."
- "Draft a post announcing [news/launch] that leads with the reader's problem, not my announcement, and only mentions the announcement in paragraph three."
- "Here's a client question I got this week: [question]. Answer it publicly as a post, as if the client is reading it, direct and specific."
- "Write three different opening lines for this post idea: [idea]. Make them different in tone, not just phrasing, one blunt, one curious, one personal."
A real example from my own feed
Last autumn I ran a small test because I was curious rather than scientific about it. Over six weeks I posted twenty times. Ten used a plain one-line prompt like "write a LinkedIn post about AI and small business." Ten used the five-part input method above, with real client details, real numbers, and a genuine opinion each time.
The plain-prompt posts averaged 34 reactions and 3 comments. The structured ones averaged 111 reactions and 19 comments, and two of them got picked up and quoted by other people in their own posts, which never happened with the generic batch. Same tool, same account, same posting times. The only variable was what I fed ChatGPT before I asked it to write.
The uncomfortable part of that experiment, the bit nobody selling a "50 ChatGPT prompts" download wants to say out loud, is that the prompt was never really the problem. My willingness to hand over an actual story, an actual number, and an actual opinion was the problem. Most people ask ChatGPT to write a LinkedIn post because they don't want to sit with the discomfort of saying something specific and risking that someone disagrees with it. A prompt cannot fix that. You have to fix that, and then the prompt just carries your voice faster.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
A step-by-step routine that takes fifteen minutes
- Open a note and write down one thing that happened this week in your business, client work, or industry. One sentence, with a number in it if possible.
- Write your gut reaction to it in one sentence, unedited, even if it sounds blunt.
- Paste both into ChatGPT with this prompt: "Here's what happened: [1]. Here's my honest reaction: [2]. Write a LinkedIn post in my voice, short paragraphs, no hashtags, ending with a specific question."
- Read the draft out loud. Anywhere it doesn't sound like something you'd say at a dinner table, rewrite that line yourself. ChatGPT gets you 70 percent there; the last 30 percent is you.
- Cut the first line ChatGPT gave you and write your own opening sentence instead. Nine times out of ten the model's opener is generic and your real first thought was better.
- Post it, then reply to every comment within the first two hours if you can, because the comment window is what LinkedIn rewards, not the post itself.
Prompts for the formats people forget
Not every LinkedIn post is a story. Here are prompts for the other formats that do well but get overlooked:
- Carousel-style text post: "Write this as a numbered list post, ten short lines, each one a standalone insight about [topic], no intro paragraph."
- Poll follow-up: "I ran a poll and 62 percent of people said [result]. Write a post reacting to that result, including why I think the other 38 percent answered differently."
- Behind-the-scenes: "Write a post that shows the messy middle of [project], not the finished result. Include one specific thing that went wrong."
- Comment-worthy question post: "Write a short post, four sentences maximum, that ends with a question people would need to think for ten seconds before answering, not a yes or no question."
If your content strategy stretches beyond LinkedIn, the same input-first approach works for other platforms too. I've written similar prompt breakdowns for Instagram caption prompts for ChatGPT and for YouTube script prompts for ChatGPT, since the core habit, feed it your specifics before you ask for the draft, carries across every platform.
What I'd tell you if we were on a call about this
If you're running a business and posting on LinkedIn is meant to bring in leads, not just likes, the prompt is the smallest part of the job. The bigger part is deciding what you think about your industry and being willing to say it plainly, then using ChatGPT to get that thought onto the page faster instead of using it to invent a thought you don't have. I see business owners spend three hours hunting for "the perfect prompt" when they'd get further in twenty minutes writing down one true sentence about their week and handing that to the model instead.
And if you're building this into a proper content system rather than doing it post by post, that's a different job, closer to setting up workflows than writing prompts, and it's worth getting outside eyes on it. That's the kind of thing an AI implementation coach would normally help you build once, so it runs without you reinventing the prompt every Monday morning.
Related guides live in the ChatGPT Prompts and Limits: 30 Guides to Fun Prompts, Photo Edits and Daily Caps. To write your own, try the free tool that writes the prompt for you.
Frequently asked questions
What is the best ChatGPT prompt for writing a LinkedIn post?
There isn't one universal best prompt. The prompts that work give ChatGPT a specific real detail, a number, and your honest opinion before asking it to draft, rather than asking it to invent a post from a topic alone.
Why do ChatGPT LinkedIn posts sound the same as everyone else's?
Because most people use the same generic prompt with no personal input, so the model defaults to the same structure: bold hook, three short paragraphs, generic closing question. Feeding it your own story and opinion is what breaks the pattern.
How long should a ChatGPT LinkedIn post be?
Most well-performing LinkedIn posts run between 100 and 250 words. Ask ChatGPT to keep drafts under 200 words and short-paragraph the output, then edit down further rather than padding it out.
Should I post ChatGPT's draft exactly as written?
No. Treat the draft as a starting point that gets you 70 percent there, then rewrite the opening line and any sentence that doesn't sound like something you would say out loud before posting it.