Write three posts for the same idea, one per platform. Each post must be platform-native, not a copy with different lengths.
The idea:
{{idea}}
Constraints (apply to all three):
- Open with the strongest line, not a setup.
- No marketing verbs: leverage, unlock, empower, transform, revolutionize.
- No openings: 'In today's fast-paced world,' 'I'm thrilled to announce,' 'Excited to share.'
- No hashtags unless I ask.
- No exclamation marks.
Per-platform rules:
- **LinkedIn**: under 220 characters before any 'see more' break, line-broken for skim, personal-voice.
- **X / Twitter**: under 280 characters, single tweet, compression-first.
- **Instagram caption**: under 125 characters, hook in first 80 chars, one CTA at end.
Social posts: LinkedIn + X + Instagram
Add to PromptPaste
Download PromptPaste for MacDownload PromptPaste for iPhoneDownload PromptPaste for iPadDownload PromptPaste for Mac & iPhone
the prompt
Example output
LinkedIn (under 220 chars, hook-first):
We almost shipped a feature nobody asked for.
Usage data was up 40% on a button we'd buried in a sub-menu.
We assumed: people love it, build more of it.
Reality: they were clicking it by mistake.
The fix took 5 minutes. The lesson took 5 weeks.
---
X / Twitter (under 280 chars):
We almost shipped a feature nobody asked for. Click data said 'usage is up 40%.' Real reason: the button was next to the one people actually wanted. The fix took 5 minutes. We almost spent a sprint expanding it instead.
---
Instagram caption (under 125 chars):
40% more clicks ≠ 40% more love. Sometimes the data is just telling you the button is in the wrong place.
Common mistakes
Don't use the same post on all three platforms - the prompt produces three because LinkedIn rewards line breaks and personal voice, X rewards compression, Instagram rewards a hook in the first 80 characters. Posting one of them on all three is worse than posting one platform-correct version. Also: do not let the model add hashtags unless you specifically ask. AI hashtag suggestions are uniformly bad and signal AI authorship louder than the words do.
More from AI Prompts for Marketers
Email subject lines (3 angles, 10 variations)
Act as a senior email marketer. Generate subject lines for the campaign below across three distinct strategic angles: curiosity, value,…
Blog outline from a target keyword
Build a blog outline for the target keyword: {{target_keyword}}. Produce, in this order: 1. The keyword's search intent (informational,…
Ad copy: 4 angles, 1 product
Write ad copy for {{product}} across four distinct strategic angles: pain, aspiration, proof, urgency. Per angle, produce one headline and…
Why it works
Generic 'write a LinkedIn post' prompts produce posts that sound like every other AI post on LinkedIn - which means they get skipped. The fix is platform-native voice plus an explicit hook structure plus a banned-words list. This prompt enforces all three. The banned-words list is the most important part: it forbids the specific verbs ('leverage,' 'unlock,' 'transform,' 'empower') and openings ('In today's fast-paced world,' 'I'm thrilled to announce') that mark a post as AI-written within two seconds. Without that list, the model defaults to them. With it, the post reads like a human wrote it.