Since 2025, one AI model reads every LinkedIn post like an editor and determines its fate. LinkedIn's 360Brew reads the text semantically, tests the post on a slice of your network for 60 to 90 minutes, and amplifies the winners to people interested in the topic for two to three weeks. As a result, organic reach fell 34% in a year, roughly 2% of the feed is company-page content, and 54% of long-form posts are likely AI-generated, which LinkedIn began suppressing in May 2026.
Attention is scarcer and half the feed sounds the same, so the structure of your post now decides who gets seen. This guide dissects the post itself: the seven zones every post is built from, the three blueprints matched to the three kinds of engagement, and the media formats that multiply or shrink your reach. Technique is learnable.
The key takeaways you need right now
- About 140 characters decide everything. Mobile collapses the caption at roughly 140 characters, and 60 to 70% of readers never click "see more." The hook carries the post.
- Engagement has a hierarchy. A save weighs about 5x a like, comment threads drive up to 2.4x reach, a 10-word comment weighs about 2x, and a like is the weakest signal on the board.
- The first 90 minutes decide the post's fate. Three or more commenters in the first hour correlates with up to 5.2x amplification. Post when you can stay.
- The carousel is the strongest format. Document posts earn 1.45x reach and about 11x the interactions of a single image. Every swipe counts as dwell time.
- Links, pitches, and bare reposts quietly kill reach. Link-built posts trail by about 42%, direct selling costs up to 75%, and a repost without commentary runs 30% below an original.
- Sounding like AI now costs distribution. LinkedIn detects generic AI text at 94% and suppresses it. Drafting with AI is fine; publishing without a fingerprint is what costs you.
- Specificity is the moat. A date, a number, and a detail only you could know beat every adjective, and they are the one thing a model cannot fake.
- The post is half the work. Slot, spacing, staying 15 minutes, and edit discipline: distribution is a routine, and consistency beats brilliance.
Part 1: The fold, where most readers stop
LinkedIn collapses every caption behind "...see more" after roughly 140 characters on mobile (desktop shows around 210). Most readers judge the whole post from that tiny little fragment. A documented A/B test on the same account, with the same audience of roughly 10,000 impressions, moved expansion from 3% to 12% by changing nothing but the first line:
Hooks rewritten in InsurTech founder voice to show the real-world application. The documented A/B tested "Let's talk about sales funnels" against "I wasted $10K on ads before I learned this": 3% vs 12% expansion, same account, roughly 10,000 impressions each.
Be mechanical: a number, a timeframe, or an outcome in the first 10 to 15 words. Justin Welsh, with 33M+ impressions behind him, says the first line is more important than 95% of the rest of the post. Test yours right here before you publish:
Test your hook against the fold
Part 2: The seven zones of the post
Every post that performs is doing seven jobs in order. Here is one example post, written the way an InsurTech founder would write it, dissected zone by zone:
She made two introductions and sat with us in both rooms. Both carriers moved from first meeting to term sheet inside eight weeks.
Zone 1, the hook (characters 1 to 140). One or two short lines carrying a specific stake: what happened, what it cost, what changed. No topic announcements, no rhetorical-question openers, no throat-clearing. If the first line works as a complete post, it works as a hook.
Zone 2, the re-hook. The reader who clicked "see more" made a micro-commitment. Reward it immediately with the second-most-interesting fact, before any background. Dwell time is an official ranking signal, and this is the line that keeps them reading.
Zone 3, the body. Short paragraphs, four lines maximum, white space between them, 800 to 1,000 characters total (roughly 100 to 160 words). Plain language wins: posts below a 10th-grade reading level earn 35%+ more reach. Vary your sentence length; uniform rhythm reads as AI.
Zone 4, the proof. At least one detail only someone who did the work could supply: a date, a real number, a named situation. "Submissions up 3x in Q1, two new capacity partners by Q3" beats any adjective, and tangible experience is the one thing a model cannot fake.
Zone 5, the payoff. One line of earned point of view, stated plainly. Skip the manufactured-reflection closers ("Let that sink in"). Teach; never pitch. Direct selling costs up to 75% of reach.
Zone 6, the ask. Optional and goal-dependent: a genuine, specific question if you want comments, nothing at all if the post stands on its own. But don't do this: "Comment YES below". It's now flagged as engagement bait.
Zone 7, the footer. A link only if necessary, always the last line: link-built posts trailed no-link posts in 34 of 37 months measured, by about 42% in 2025. Tag one or two people you genuinely expect to reply. Zero to three niche hashtags; six or more actively hurts.
Part 3: Engagement has a hierarchy
All engagement is not equal. The feed weighs what a reader does with your post, and the spread between the strongest and weakest signals is roughly five to one:
Every one of those signals is on a clock. A post's fate is decided in a test window shorter than a lunch break:
Post when you can hang out, watch, and engage right away, because author replies are fuel. And strong posts now resurface for two to three weeks, up from a 24-hour shelf life, so the window pays compound interest. Decide which signal you are building for before you draft, because the three targets produce three different posts.
Part 4: Three blueprints, one per signal
THE REFERENCE
THE CONVERSATION
THE MOMENT
The Reference: posts built to earn saves
The reader saves what they expect to need again, so package reference value: checklists, planning calendars, step-by-step processes, benchmark tables, mini-guides. The natural format is the document carousel, the strongest format on the platform. Structure the carousel like the post itself: the cover slide is the hook and obeys the same 140-character discipline, each interior slide carries one idea, and the closer holds the takeaway and any URL. In the caption, tell the reader what they will want later ("the planning calendar we walk clients through every spring") and let the save invite itself. Evergreen beats topical here, because saves compound across the two-to-three-week resurfacing window.
The Conversation: posts built to earn comments
Comments in the golden window decide amplification, so draft for discussion and then stay in it:
- Take a position someone could reasonably push back on, held with evidence. Defensible tension draws replies; consensus draws likes.
- End with one specific, answerable question tied to the reader's own experience. A generic "Thoughts?" does nothing.
- Leave something incomplete on purpose. The post covers your approach; the comments collect everyone else's.
- Post only when you can stay 15 minutes. Author replies build the threads that drive up to 2.4x reach and can lift distribution about 40%.
- Tag the one or two people whose answer you actually want. A wall of silent tags reads as spam.
The Moment: posts built for reactions
Reactions are the weakest ranking signal, so use this blueprint deliberately: recognition, promotions, anniversaries, events, and team wins, where broad low-friction visibility is the point. Lead with an image of real people, which earns up to 50% more engagement than stock art. Keep the caption short and warm; the image earns the stop and the caption tells your audience why it matters. Congratulate people by name, because their networks carry the post into rooms yours cannot reach.
Part 5: Rich media, and when to use it
Format multiplies or shrinks everything above. From a 621,833-post dataset on personal profiles:
- Carousel: for processes, checklists, guides, and data stories, anything save-worthy. Skip it when the idea fits in one image; padded slides kill the swipe-through that makes the format work.
- Image with real people: for human moments, recognition, events, and one strong data visual. Skip it when all you have is stock art; a generic graphic adds nothing a text post lacks.
- Native video under 60 seconds: for personality and conviction on camera, demos, and event energy. Vertical, mobile-first, watchable with sound off. Upload native, never a YouTube link. This is LinkedIn's strategic push, with viewership up 36% year over year, so expect the multiplier to keep improving.
- Text-only: for a story strong enough to carry itself, and the fastest to ship. Pair it with a visual when one exists.
Part 6: What will kill your post
Most underperforming posts trace to one of four penalties:
The rest of the kill list: posting twice within 24 hours throttles the first post mid-test. Engagement bait and coordinated pods are demoted. Coordinated pods are groups of people you message privately to guarantee likes and comments on each other's posts, and LinkedIn began restricting pod accounts and automated comments in February 2026. Hashtag stuffing hurts (the pages were retired in 2024; zero to three niche tags at most). And generic AI text is suppressed at 94% detection.
Draft with AI if you like, then add the detail only you know before you publish.
Part 7: Publish, then stay
The post is half the work. The first hour is the other half.
- Slot: Tuesday to Thursday, mid-morning to midday, then trust your own analytics. Eight million analyzed posts agree on the days; the hour varies by study.
- Spacing: never two posts within 24 hours. Cadence of two to three posts a week is the sustainable start; three to five is the data's sweet spot.
- Stay 15 minutes: reply to every early comment. The 60-to-90-minute window decides the post's fate, and scheduling tools are fine only if you show up for it.
- Edit discipline: fix typos immediately, and never restructure or add links to a live post.
- Lanes: keep to your two or three topics. The model matches consistent authors to interested audiences, and lane-hopping resets what it knows about you.
Score your next post
Twelve questions, one point per yes. Run the draft before you publish. Be brutal.
The pre-publish scorecard
Structure is learnable. The voice on top of it is the asset.
RevUp Studio is a builder studio for InsurTech and B2B tech founders. We build brand, content systems, and AI-native marketing engines the founder owns outright, and founder voice is where every engine starts. If you want a second set of eyes on how your team shows up on LinkedIn, we read every reply.
Appendix: where the numbers come from
Corpus. This reference distills a 30-source review compiled for a July 2026 team training. The large datasets: Richard van der Blom's Algorithm Insights Report (1.8M posts, 58K profiles), AuthoredUp's algorithm study (621,833 posts), Metricool's LinkedIn study (673,658 posts), Ordinal's link-penalty study (900K+ posts over three years), and Socialinsider's 2026 benchmarks (1.3M posts). Platform-official items: LinkedIn's 360Brew model paper, its May 2026 AI-content suppression announcement, its February 2026 pod restrictions, and its statement on links.
A note on data quality. LinkedIn does not publish its ranking weights. Official items are platform statements or LinkedIn-published research; the third-party datasets are observational and directionally reliable rather than exact. Figures widely repeated online but untraceable to an original study (the "561% reach" advocacy claim, "comments are worth 15x likes") were deliberately excluded.