On June 30, LinkedIn did something platforms almost never do: it published the actual rules for getting picked up by AI search — and the rules have almost nothing to do with how big your audience is.

Some background on why this matters. When someone asks ChatGPT, Perplexity, or Google's AI Mode a business question, the answer cites a handful of sources. Semrush analyzed 89,000 of those citations earlier this year and found LinkedIn shows up in roughly 11% of AI answers — ahead of Wikipedia, ahead of YouTube, ahead of every major news publisher. If you post there, your writing is now raw material for AI answers, whether you planned for it or not.

Here's the number that should change how you post. In that same dataset, the median LinkedIn post cited by AI had 15 to 25 reactions and at most one comment. One article — 31 likes, 12 comments — ranked among the most-cited URLs in the entire study, appearing in 45 separate ChatGPT answers. Meanwhile, plenty of viral posts with thousands of reactions were never cited at all.

The finding

The AI never checks your follower count. It checks whether your post answers the question cleanly — and most viral posts don't answer anything.

What actually gets a post quoted

LinkedIn's guide reads more like a technical spec than social media advice. The parts worth acting on:

Your first line becomes your URL — permanently. LinkedIn generates a post's web address from its opening sentence the moment you hit publish, and editing the post later doesn't change it. Open with "#mondaymotivation" and you've locked in a junk URL forever. Open with the actual topic — "How to structure a client report" — and the address itself tells every retrieval system what the post is about. One decision, no do-overs.

Posts and articles do different jobs. Short posts (200–300 words is the sweet spot) get cited for narrow, specific questions. But long-form articles of 800–1,200 words earn 50 to 66% of all LinkedIn AI citations, because they have the headings, structure, and depth a model can pull a complete answer from. If you've gone all-in on short posts because that's what the feed rewards, the AI layer rewards the opposite.

Structure is the ranking factor. Among the most-cited articles: 100% used bullets or lists, 92% used clear headings, two-thirds included real numbers. Add dates to your text so the AI can tell your information is current. Write the question, then answer it. Boring, legible structure — the stuff that never goes viral — is exactly what gets retrieved.

Original beats recycled. About 95% of citations go to original content, not reshares or commentary on someone else's post. And the workable cadence LinkedIn recommends is modest: one article a week, broken into two or three short posts that each carry one idea from it.

Why this is worth your Tuesday morning

Because the old channel is shrinking underneath you. Ahrefs measured it in February: when Google shows an AI Overview, clicks to the top-ranked result drop by 58%. Fewer people are clicking through to anyone's website. The visibility that's replacing those clicks is being named inside the answer — and you can't buy that placement. Paying for ads next to AI answers doesn't get you cited in them.

For a small business, this is unusually good news. The viral game demands an audience you may not have. The citation game demands structure, which is free. One well-organized article a week — a real question your customers ask, answered with headings, numbers, and a date — is a realistic output for one person, especially with AI drafting alongside you. The post that wins isn't the loudest one. It's the one the machine can quote.