35 AI Prompts for LinkedIn Posts That Get Engagement

TL;DR: These 35 linkedin ai prompts are organized by framework rather than by topic: personal story and contrarian take first, because they carry most B2B posts that actually get read, then five supporting categories — results, frameworks, commentary, career reflection, and questions. Every prompt is a copy-paste block with fields you fill in, and none of the engagement numbers below are invented.

Why do most AI-written LinkedIn posts get scrolled past?

Because most of them are answers to a topic instead of a position on one. "Write a LinkedIn post about remote work" produces a paragraph restating what everyone already believes about remote work, phrased slightly differently than the last version. It is not wrong, and it is not read, because it asks nothing of the reader and risks nothing for the writer.

LinkedIn is a professional network with a business audience and real buying power behind a meaningful share of the accounts scrolling it, which is exactly why generic advice performs worse there than it does on a platform built for entertainment. A reader evaluating whether to work with you, hire you, or buy from you is reading for a signal about how you think, not for a restatement of consensus.

The fix is not a better sentence. It is a different instruction to the model: ask for a structure that takes a position, not a paragraph about a subject. Two structures do most of that work, which is why they get the first ten prompts and the deepest treatment below. The other 25 prompts, organized into five more categories, round out a month of posting once the two anchor frameworks are in your rotation.

This is also why a topic-first prompt list underperforms a framework-first one even when both are run through the same model. "Write about X" leaves the structural decisions, where the post opens, what it argues, what it asks of the reader, entirely up to whatever pattern the model defaults to, which tends toward the safest, most averaged version of a LinkedIn post it has seen. Fixing the structure first and leaving only the specific content to fill in produces a narrower, more specific result every time, because the model is no longer solving two problems (what to say and how to say it) in one pass.

The personal-story prompt: the framework that carries most B2B posts that actually get read

A personal story works on LinkedIn because it is the one thing a competitor cannot copy. Anyone can restate advice. Only you have the specific afternoon a client rejected your proposal for a reason you did not see coming.

ROLE
You are a first-person LinkedIn writer, not a marketing copywriter. You
write the way a specific person talks, not the way a brand talks.

INPUT
What happened: {{THE_EVENT}} — a real, specific moment, not a summary.
When and where: {{WHEN_WHERE}}
What I got wrong, or what I did not expect: {{THE_TURN}}
What I do differently now because of it: {{THE_LESSON}}
Audience who should see themselves in this: {{AUDIENCE}}

TASK
Write a LinkedIn post that opens with the moment itself, not with "I want to
share a story" or any sentence that announces what kind of post this is.
Structure: opening moment (2-3 lines), what went wrong or surprised me
(1 paragraph), the turn (1-2 lines), the lesson stated plainly, one line
inviting a specific kind of response from {{AUDIENCE}}.

RULES
- No sentence that states the moral before the story finishes.
- No "game-changer," "unlock," or forced enthusiasm anywhere.
- Keep it under 1,300 characters. If it runs long, cut context, not the
  specific moment.
- Do not invent a number, a client name, or a detail I did not give you.

OUTPUT
The post. Then, separately, three alternate opening lines using the same
event, so I can pick the one that sounds most like me.

The "three alternate opening lines" instruction matters more than it looks like it should. A model's first draft of an opening line tends toward the safest, most generic phrasing it can generate. Asking for three forces it to actually vary the angle, and the version that sounds least like a LinkedIn post is usually the one that reads best on LinkedIn.

Four more prompts in the same framework, each pointed at a different kind of moment:

1. THE MISTAKE POST
Event: a specific decision that went wrong. Ask for: what I believed
going in, what actually happened, what I would tell myself before making
that call again. Ban any sentence that excuses the mistake.

2. THE FIRST-TIME POST
Event: the first time I did {{THE_THING}} — the first sale, the first
hire, the first time a plan fell apart in front of a client.
Ask for: what I did not know that everyone assumes you know already.

3. THE CHANGED-MY-MIND POST
Event: something I used to believe about {{TOPIC}} and no longer do.
Ask for: what changed it — a specific conversation or result, not a
gradual realization. State the old belief before the new one.

4. THE SMALL-DECISION POST
Event: one specific, unglamorous decision that had outsized results.
Ask for: the decision, why it looked unimportant at the time, what it
actually produced. Explicitly avoid making it sound like a strategy.

The contrarian-take prompt: naming the belief before you disagree with it

A contrarian take only works if it names the specific, common belief it is arguing against before disagreeing with it. Skip that step and the post reads as an opinion with no target, which is much easier to scroll past than an opinion that picks a fight with something the reader actually believes.

ROLE
You are someone with direct, defensible experience in {{FIELD}}, writing a
post that disagrees with a common belief in that field.

INPUT
The common belief, stated the way people who hold it would state it:
{{COMMON_BELIEF}}
What I actually think instead: {{MY_POSITION}}
Why — the specific reasoning or experience behind it, not just an opinion:
{{REASONING}}
Where my position has a real limit or does not apply: {{HONEST_LIMIT}}

TASK
Write a LinkedIn post that states {{COMMON_BELIEF}} in its strongest, most
reasonable-sounding form first — not a strawman version. Then disagree with
it directly, using {{REASONING}}. Close with {{HONEST_LIMIT}} stated
plainly, not softened into a disclaimer.

RULES
- The common belief must be stated fairly. A weak strawman makes the
  disagreement look easy instead of earned.
- No "hot take:" or "unpopular opinion:" labels — let the content signal
  that on its own.
- Include the honest limit even if it weakens the post. A position with
  no stated limit reads as either naive or dishonest to an experienced
  audience.
- Under 1,500 characters.

OUTPUT
The post, then one sentence naming who would disagree with {{MY_POSITION}}
and why they are not wrong to.

The honest-limit field is the one most contrarian-take prompts skip, and it is the reason so many "unpopular opinion" LinkedIn posts read as performance rather than conviction. A position stated with no acknowledged edge case is a position nobody has actually tested against reality yet.

Four more contrarian prompts, aimed at different targets for the disagreement:

1. THE METRIC POST
Common belief: {{METRIC}} is the number that matters most in {{FIELD}}.
Position: it is not, or not alone. Reasoning: what it hides or what it
does not measure. Honest limit: where {{METRIC}} genuinely is the right
call.

2. THE TOOL/PROCESS POST
Common belief: everyone in {{FIELD}} should use {{TOOL_OR_PROCESS}}.
Position: state when it is the wrong choice. Reasoning: a specific
situation where it failed or added no value. Honest limit: who should
still use it.

3. THE ADVICE-YOU'VE-HEARD POST
Common belief: a specific piece of advice people in {{AUDIENCE}} repeat
constantly. Position: it is incomplete or situational, not universal.
Reasoning: the case where following it made things worse.

4. THE HIRING/CAREER POST
Common belief: a common assumption about what makes someone hireable or
promotable in {{FIELD}}. Position: name the thing that actually predicted
it in your experience. Reasoning: a specific case, not a generality.

Twenty-five more prompts, organized by what the post is actually trying to do

Personal story and contrarian take carry most of the weight, but a month of only those two starts to feel like a performance of vulnerability and disagreement rather than a real account. These five categories round out a rotation without repeating either framework every time.

Results and case study (5 prompts)

These work when you have a real, specific outcome to point at — a number, a before-and-after, a project you can name without breaching confidentiality.

1. Result: {{OUTCOME}}, achieved by {{WHAT_WE_DID}}. Write a post that
   leads with the outcome in the first line, explains the two or three
   decisions that produced it, and names one thing that almost didn't
   work.

2. A before-and-after for {{CLIENT_OR_PROJECT}} (anonymized if needed).
   Structure: the state before, the specific change made, the state
   after. No vague "significant improvement" language — use the real
   number or say [NO NUMBER AVAILABLE] instead of inventing one.

3. A project that failed to hit its goal. State the goal, what actually
   happened, and the one decision you'd change. Ban any sentence that
   reframes the failure as secretly a win.

4. A comparison of two approaches to {{PROBLEM}} we tried, with the
   result of each. Structure as a short story, not a bulleted
   comparison table.

5. A client or customer quote, used verbatim from {{QUOTE_SOURCE}}, with
   one paragraph of context around it explaining what problem it solved.

Frameworks and how-to posts (5 prompts)

These earn a save or a "sent this to my team" comment when the process described is one you actually use, not a generic best-practice list assembled from what a model thinks a process should contain. Ask for the real steps you follow, in the order you actually do them, not the order that sounds most logical after the fact.

1. A process I use for {{TASK}}, as a numbered list of 3-5 steps, each
   one sentence. No step should require a sub-list.

2. The one question I ask before {{DECISION_TYPE}}, and why that
   question specifically, not a generic checklist.

3. Three signs that {{PROBLEM}} is about to happen, based on
   {{REAL_EXPERIENCE}}, written as a warning post.

4. A comparison of what {{AUDIENCE}} thinks {{TASK}} requires versus
   what it actually requires, in two short columns of text (not a
   table — LinkedIn strips table formatting).

5. A framework named after its own structure (e.g., "the three
   questions," "the one-line test") built from {{REAL_METHOD}}, not
   invented for the post.

Industry commentary and trend reaction (5 prompts)

This category ages the fastest of all seven, which is exactly its advantage: a post reacting to something that happened this week reaches people who are already thinking about it, and reaches almost nobody a month later. Write these close to the event itself, and resist the urge to make the take broader or more timeless than the moment actually calls for.

1. React to {{RECENT_DEVELOPMENT}} in {{FIELD}}: what most people are
   saying about it, what I think they're missing, and why it matters
   for {{AUDIENCE}} specifically.

2. Explain {{RECENT_DEVELOPMENT}} to someone in {{AUDIENCE}} who has
   heard the term but not the substance, in under 150 words.

3. Predict one concrete change {{RECENT_DEVELOPMENT}} will force in how
   {{AUDIENCE}} works, stated as a specific claim I'd be willing to be
   wrong about.

4. Compare {{RECENT_DEVELOPMENT}} to a past shift in {{FIELD}} that
   {{AUDIENCE}} lived through, and say what actually changed versus
   what people predicted at the time.

5. State the boring, unglamorous implication of {{RECENT_DEVELOPMENT}}
   that nobody is posting about, because it is not exciting enough —
   only if I actually believe it, not as a contrarian exercise.

Career reflection and lessons (5 prompts)

The failure mode here is writing from the position of having already arrived somewhere, which reads as a humblebrag no matter how the sentence is worded. The stronger version names a specific point of confusion or a specific wrong assumption you actually held, and lets the reader recognize themselves in the confusion rather than in the conclusion.

1. The advice I'd give someone starting {{ROLE}} today, based on
   {{YEARS}} years doing it, limited to the one thing I'd say if I only
   had one sentence.

2. A skill that mattered less than I expected when I started {{ROLE}},
   and what mattered more instead.

3. The moment I realized {{ROLE}} required something different than I
   was trained for, told as a specific scene, not a generality.

4. What I stopped doing in {{ROLE}} that I used to think was essential,
   and what replaced it.

5. A question I wish someone had asked me before I took {{DECISION}},
   posed to the reader directly at the end of the post.

Questions and discussion starters (5 prompts)

A question post only works if you would genuinely want to read the answers. A question asked purely to generate comments reads as exactly that within a sentence or two, and an audience that has been asked "agree or disagree?" enough times stops answering. The prompts below all require you to supply a real stance or a real uncertainty first, not just a topic.

1. A genuine either/or question about {{TOPIC}} where I actually don't
   know the more common answer, framed as a LinkedIn poll with two to
   four options.

2. "What's the {{SUPERLATIVE}} you've dealt with in {{FIELD}}?" — with
   my own one-paragraph answer posted first, so the question isn't
   asked from a position of having nothing to say myself.

3. A fill-in-the-blank prompt about {{TOPIC}} where the blank has
   several defensible answers, not one obvious correct one.

4. A direct question to {{AUDIENCE}} about a decision they're currently
   facing, with context on why I'm asking (not just curiosity-bait).

5. "Am I the only one who thinks {{POSITION}}?" — used only when
   {{POSITION}} is a real, specific opinion I hold, not a manufactured
   provocation.

What LinkedIn actually limits, and what these prompts do not promise

LinkedIn's own Help Center states plainly, checked 25 September 2026: "The character limit for a post is 3,000 characters." Most of the prompts above ask for something well under that on purpose, closer to 1,300–1,500 characters, because a post that needs the "see more" click to reach its point has already lost some of the readers who would have engaged with a shorter one.

None of the 35 prompts above, and nothing in this post, claims a specific engagement lift for any format, framework, or posting time. Any number you see elsewhere claiming a fixed percentage increase from a poll, a question, a specific posting hour, or a hashtag count did not come from LinkedIn's own published data as far as we could verify, and we are not going to manufacture one to make this page sound more authoritative than it is. What we can say is mechanism, not measurement: a poll asks for a response, a specific detail is harder to scroll past than a generic one, and a stated position gets more comments than a restated consensus. Test each framework against your own account and trust that result over any blog's number, this one included.

The same caution applies to hashtags and posting time, both of which get treated as settled science in most LinkedIn advice. Neither is something we have measured or can point to a LinkedIn source for, so the honest answer is: pick a small number of hashtags that actually describe the post's subject rather than a maximum-reach guess, and post at a time you can be consistent with rather than a specific hour some other account swears by. Consistency you can sustain beats a schedule copied from someone else's result.

If your calendar is planning LinkedIn alongside other platforms rather than LinkedIn alone, the social media calendar generator handles the cross-platform rotation and pillar planning; this page goes deeper on the two frameworks that specifically carry LinkedIn. If these posts are drawn from a longer piece of writing rather than built from scratch each time, the content repurposing workflow covers turning one article into a week of angles. And if the follow-up conversation that starts in a LinkedIn comment moves to email, 30 AI prompts for follow-up emails that get replies picks up exactly where this page leaves off.

Where Prompt Architects actually fits

Prompt Architects does not post to LinkedIn, does not schedule anything, and does not measure engagement on your account. What it does is turn the two framework prompts above, and the 25 that follow them, into reusable templates with {{COMMON_BELIEF}}, {{THE_EVENT}} and {{FIELD}} stored as Global Variables you fill in per post instead of rewriting the scaffolding every time. Keeping your brand voice consistent across 35 different framework prompts is exactly what a saved brand-voice context is built for, so a contrarian-take post and a personal-story post still sound like the same person wrote them.

There is a free plan, checked on the live pricing page 25 September 2026, that includes 5 prompt generations a day with no card required. Current pricing for saving these as a reusable library is on the pricing page.