Marketing
Why some LinkedIn posts get saved (and most don’t)
Most LinkedIn posts don’t fail to go viral. They fail to earn dwell, saves and replies. How to write for those three, without a viral promise.
lekterable6 min read
Summarize withChatGPTClaudePerplexityGoogle AI Mode
Short answer: posts spread when people stay on them, save them and reply to them. Do not write for viral. Viral is a lottery ticket other people screenshot. You cannot run a week on lottery tickets.
What you can run a week on: dwell (someone stayed), saves (they want it later), replies (they had something to add). LinkedIn redistributes posts that earn those. A post with 27 views is not a moral failure. It is a post that did not earn a second look from the people who saw it.
This page is the difference between a post that sits and a post that circulates a bit — without promising it will blow up.
The system around the post is the LinkedIn content system. Reach context (without “beat the algorithm” magic) is LinkedIn organic reach. If the draft sounds like a model, start from why ChatGPT posts suck.
The three signals that are not vanity
Dwell. Did a person stop and actually read? First two lines earn the rest. A wall of text, or a hook that could sit on anyone’s profile, does not get dwell. You will not see a perfect dwell metric in the native UI. You will see the pattern: posts people finish get more distribution than posts people bounce.
Saves. Did someone keep it because it was useful later — a checklist, a sentence they want to steal the structure of, a lesson they will need on Thursday? Saves are a stronger “this mattered” than a reaction emoji.
Replies. A real comment, especially in the first hour, is how a post keeps circulating. “Great post” is not a reply. A disagreement, an example, a “here is how this failed for us” is.
Likes without those three are greetings. Do not manage the week to greetings.
Why your “brilliant” post sat there
Usually one of these, not the algorithm having a mood:
- The first two lines were about you. Job titles, “I’m excited to share,” “In today’s fast-paced world.” A stranger has no reason to stay.
- There was no scene. Tips without a moment are interchangeable. Transformation (“what changed”) is a specific before and after, not the word “transformation.”
- It was a block. Huge paragraphs, no air, no “see more” worth opening. Write for a phone thumb, not an essay PDF.
- The idea was not yours. A prompt that starts “write a viral LinkedIn post” produces a post that could belong to anyone. Anyone’s post does not get saved.
- You left after you hit publish. No replies in the first hour, so the only people who saw it were the first few in the graph, and they scrolled.
None of this is a secret. None of it is a guarantee.
Write for dwell, then for a save, then for a reply
Dwell: a specific claim or scene in line one and two. A number you actually have, a mistake you actually made, a sentence a customer actually said (with permission). Not “Nobody talks about this.” Everybody talks about that.
Save: one thing a reader can use without you. A three-step shape. A rule for missed days. A before/after of a hook. If they cannot use it on Thursday, they will not save it.
Reply: ask for an opinion or an example, not a favor. “Which of these three pillars would you kill?” works. “Please comment for the algorithm” does not.
Then stay. Reply to every comment on your own post, especially early. That is distribution you do not have to buy. It is also the loop: a good comment is next week’s inbox line.
Structure that does not kill the post
- Short lines. Short paragraphs. One idea per sentence.
- Hook → story or insight → takeaway → soft CTA.
- Preview it as a LinkedIn post with a body — line breaks, “see more,” the card — before you publish. The free LinkedIn post preview is built for this. A Google Doc or a Buffer-style queue hides the thing people actually see.
Paste the draft into that preview. If you would not stop for it, do not schedule it.
What not to do
- Do not prompt a model to “write a viral post.”
- Do not invent a percentage of reach you “lose in the first two seconds.”
- Do not chase someone else’s one-liner. You do not have their graph or their week.
- Do not post more to compensate. Post the same three pillars, in a cadence you keep.
This week
- Take one ignored post. Rewrite only the first two lines so a stranger would care.
- Add one usable takeaway (the save).
- End with one opinion question (the reply).
- Preview with a body. Ship. Stay for the first hour.
If you want help holding the shape, generate a draft from a real inbox line — a call, a quote, a miss — not from “viral.” Free plan: 5 AI posts a month, no credit card.
Questions people type next
Why did a simple one-liner get more reach than my useful post?
Often: a tighter first line, a denser graph, or replies you cannot see from the outside. Copying the one-liner without the graph does not transfer the result. Write for your pillars.
Can I make a post go viral on LinkedIn?
You can make a post easier to stay on, save, and reply to. You cannot honestly promise viral. Anyone selling that is selling a story.
Is emotion “the real algorithm”?
Emotion helps a stranger recognize themselves. It is not a switch. A feeling with no takeaway still dies in the feed.
Ready to write for saves, not lottery tickets?
You do not need a scroll-stopping formula. You need a job, three pillars, a shape, and the patience to earn dwell.
Generate a draft from something that happened this week. Preview it with a body. Then ship and stay for the replies.
Related posts
Keep going with the LinkedIn content system.
How to get more LinkedIn engagement (without hacks)
More LinkedIn engagement means dwell, saves and replies. They come from structure, a human voice and staying for the first hour, not a promised percentage.
LinkedIn content calendar: cadence you can keep
A LinkedIn content calendar is a weekly batch you can keep: three pillars, three to five drafts, scheduled or parked, and a rule for missed days.
Why your ChatGPT posts suck (and how to fix them)
ChatGPT posts sound generic when the prompt has nothing in it. A before and after on the same topic, and the three-part prompt that fixes it.