How to Prompt AI for Better LinkedIn Posts
Generic LinkedIn posts get ignored. The prompt is almost always the problem — not the AI. Here's the specific change that turns AI-written LinkedIn content from forgettable to shareable.
Write a LinkedIn post about leadership.
Why it underperforms:
- ✗Topic is too broad — 'leadership' produces advice anyone could have written
- ✗No personal angle — LinkedIn performs best with first-person observation
- ✗No hook structure — AI defaults to starting with the conclusion
- ✗No target reader — the post won't speak to anyone specifically
- ✗No format guidance — AI may write a block paragraph instead of a scroll-stopping format
Write a LinkedIn post from the perspective of an experienced engineering manager sharing a lesson about why their first instinct as a new manager — to fix problems immediately — was damaging their team. First-person voice throughout. Open with a one-sentence hook that creates tension or surprise (do not start with 'I'). Use short paragraphs of 1-2 sentences. End with a direct question to prompt comments. 150-200 words total. No hashtags in the body.
What changed:
- ✓Specific perspective and lesson — something observed, not generic advice
- ✓First-person voice matches LinkedIn's native performing format
- ✓Hook instruction prevents the AI from burying the lead
- ✓Short paragraph format is readable on mobile and matches what performs on the platform
- ✓Call to comment increases engagement signals that expand reach
Output Comparison
Output from weak prompt
Leadership is one of the most important skills in any organization. Great leaders inspire their teams, communicate clearly, and lead by example. Whether you are a new manager or a seasoned executive, there are always opportunities to improve your leadership skills and make a greater impact on the people around you.
Output from improved prompt
The fastest way I damaged my team's trust was by being too helpful. When I became an engineering manager, I thought my job was to solve problems fast. Someone hit a blocker? I'd jump in. A deadline at risk? I'd take over. I thought I was being supportive. I was actually sending a clear message: I don't think you can handle this. The turnaround came when a senior engineer told me: 'Every time you step in, I feel like I failed.' That conversation changed how I managed. I started asking questions instead of giving answers. I let people sit with the discomfort of a hard problem. The team got stronger. Problems got solved faster — by them, not me. New managers: what's the hardest habit you've had to break?
Why It Works
LinkedIn's algorithm rewards posts that generate comments in the first 30-60 minutes. The hook formula (tension/surprise in sentence one), short paragraphs, and a closing question all serve this goal. The specific perspective (new manager making a specific mistake) gives the AI a concrete story to tell rather than generic advice. The 'do not start with I' instruction forces a stronger opening.
The Technique: Platform Context + Hook Formula + Audience Mirror
Platform Context tells the model what format norms apply to the specific distribution channel. Audience Mirror means writing from a specific perspective your reader recognises — 'experienced engineering manager' resonates with both people in that role and people who report to one. Hook Formula gives the AI a structural rule for sentence one, preventing the most common LinkedIn AI failure: starting with the conclusion.
Next step: use it in Claude Code
Prompts like this one are most useful when they are pinned into a CLAUDE.md or wrapped in a slash command. The Claude Code guide shows you how.
Read the Claude Code guideFrequently Asked Questions
Does this work for other platforms like Twitter/X or Instagram?
Yes, but you need to replace the platform-specific instructions. For X/Twitter, specify thread vs single post, character limits, and whether to use line breaks. For Instagram, specify caption length and whether to put hashtags in the caption or comment.
Why does telling the AI not to start with 'I' improve the hook?
Starting with 'I' on LinkedIn is extremely common and the eye skips past it. A hook that opens with a statement, observation, or provocative claim creates more pattern interruption. The constraint forces the AI to find a stronger entry point.
Should I include my actual experience in the prompt?
Yes — the more specific the story you give the AI, the better. Instead of 'an engineering manager' you could write 'an engineering manager at a 40-person SaaS company who moved from IC to manager six months ago.' Real detail produces authentic-sounding output.