ChatGPT Prompts for LinkedIn Posts

ChatGPT prompts for LinkedIn work when you feed a real story, a named reader, and a ban on invented metrics. You leave with ROLE / TASK / FORMAT wrappers for feed posts that hook before "see more," comments that add one useful point, and About-section drafts that fit LinkedIn's 2,600-character cap as of mid-2026. Paste facts you can defend if a commenter asks a follow-up. GPT-5.5 Instant handles first drafts. GPT-5.6 handles voice and honesty audits. This page covers LinkedIn. Multi-platform social templates live in a separate guide. Soft tip: PromptMake /text can turn a rough idea into a structured scaffold at https://promptmake.net/text before you paste into ChatGPT.

Who chatgpt prompts for LinkedIn help

You write on LinkedIn and want ChatGPT to shape a hook, a comment, or an About draft while each claim stays interview-defensible. The patterns fit operators who freeze on the first line, job seekers who need an About rewrite that matches a resume they own, and specialists who comment on other people's posts and want a useful sentence instead of "Great share." Founders who ship a product update and need a post from notes land here too.

Skip this page if you want one prompt that also covers X, Instagram, TikTok, and Facebook. That multi-platform FORMAT overlay lives in the social media templates guide. Skip it if you want resume bullets or a cover letter. Those jobs have their own prompts. This article stays on LinkedIn surfaces: the feed post, the comment, and the About section. Headline rewrites under 220 characters can reuse the same fact sheet in a second pass.

Treat ChatGPT as a draft partner with constraints. You supply the story, the reader, the surface, and banned claims. The model proposes line breaks, a hook that fits the fold, and verbs that sound like you. You reject anything you cannot defend if a recruiter or a former colleague replies in public.

Core prompt pattern for LinkedIn writing

Strong chatgpt prompts for LinkedIn give four inputs before any tone request: the story facts, the reader, the surface, and the honesty boundary. Story facts are the incident, the lesson, the numbers you own, and the role you hold. The reader is the job title or industry you want to reach, plus the action you want after they read. The surface names post, comment, or About. The honesty boundary forbids invented metrics, fake client logos, fake job titles, quotes you did not say, and awards you did not receive.

Paste those four blocks near the top. Put format rules next. Put banned phrases at the end so they survive a long paste. GPT-5.5 Instant and GPT-5.6 follow labeled blocks. A vague "write a LinkedIn post about leadership" prompt produces generic fluff because ChatGPT has no story, no reader, and no forbidden list. Aim for one surface per thread. Mixing a feed post and an About rewrite in the same chat muddies length and hook rules. Run a second pass if you need the other surface.

Keep a LINKEDIN_FACT sheet outside ChatGPT: current title, who you help, two proof points, tools you use, and any metric you can defend. Update that sheet when you ship a project. Feed the sheet into each prompt. ChatGPT should not be your memory. Store a short voice sample too: two posts you published that sound like you. Paste them as SAMPLE_A and SAMPLE_B so the model copies cadence instead of LinkedIn-guru cadence.

Role, task, format skeleton

Copy this skeleton and fill the brackets with your material:

ROLE: You are a LinkedIn writing editor for [your title and industry]. You rewrite from my facts. You never invent employers, clients, titles, dates, metrics, awards, or quotes.

TASK: Turn STORY_FACTS into a [feed post | comment | About section] for READER. Match the voice in SAMPLE_A. Keep every claim tied to STORY_FACTS. Mark gaps with [NEED FACT].

FORMAT: Follow OUTPUT_SHAPE. Short sentences. Line break after every one or two sentences for posts. No hashtag in the hook. Cap length as listed. If I gave no number, write no number.

READER: [job title or industry, and the action I want: comment, connect, visit URL, or none]

STORY_FACTS: [paste notes: what happened, what you did, proof you own, lesson in one line]

SAMPLE_A: [paste a post you wrote that sounds like you]

OUTPUT_SHAPE: [hook / body / close for a post; 2 to 4 sentences for a comment; hook + proof + CTA for About]

RULES: No "excited to announce" or "thrilled to share" unless I wrote those words. No fake humility. No invented namedrop. Flag any line that needs a missing fact with [NEED FACT] instead of guessing.

REMINDER: Never invent metrics or client names. Use [NEED FACT] for gaps.

That last reminder line stops confident fake percentages. Models love tidy numbers. Your rule forces a gap marker you can fill later from a dashboard, a contract, or memory you trust.

Voice and honesty rules

Voice fails when you ask for "a thought-leadership post" with no sample. Name the cadence: first person, short lines, one idea. Paste SAMPLE_A. Ask ChatGPT to list three traits of that sample, then write the new piece with those traits. Run Instant for that trait list. Keep the list in the next message so the draft stays tied to it.

Honesty fails when you paste a resume and say "make me sound senior." ChatGPT will inflate scope. Ban title inflation in RULES. If you were an IC, the post cannot call you a Head of. If you supported a launch, the post cannot say you owned the P&L unless STORY_FACTS say you did.

Banned-phrase list that earns its keep: "game-changer," "leverage," "synergy," "humbled to," "on a journey," and "10x." Add your own cliches. Put the list in RULES. Ask for a second pass that cuts any sentence a stranger could paste onto another profile without changing a noun.

Draft each LinkedIn surface with a dedicated prompt

You lose readers in three LinkedIn boxes if you paste one blob of copy. Readers tap "see more" after the first 140 to 210 characters of a feed post, and the hard cap sits near 3,000 characters as of mid-2026. A comment has room up to about 1,250 characters, and readers see the whole thing, so a long essay in someone else's thread reads like a hijack. The About section allows about 2,600 characters, with the first 230 to 350 visible on a profile before a tap. You need a FORMAT block per surface because hook length, proof placement, and close type change. Keep STORY_FACTS stable. Swap TASK and OUTPUT_SHAPE. Do not ask ChatGPT for "a LinkedIn version" without naming which box on the site you will paste into.

Company-page posts can reuse the same skeleton if you change ROLE to the page voice and swap first person for "we" when the fact sheet supports it. Personal-brand posts stay in first person. Mix those in one thread and you get a hybrid that sounds like a press release.

Feed posts that survive the see-more cut

Put the hook in FORMAT as a hard character budget: max 200 characters, complete sentence, no hashtag, no emoji stack. Ask ChatGPT to print the hook character count on its own line so you can check before you paste. Body: 3 to 6 short paragraphs, one idea each, proof from STORY_FACTS, then one close. Close is a question or a single next step. Hashtags: 3 at the end from an approved list you paste, or zero if you do not use them.

Post skeleton extra lines:

OUTPUT_SHAPE: Line 1 to 2 = HOOK (max 200 characters). Then BODY (target 900 to 1,600 characters unless I raise the cap). Then CLOSE (one question or one URL). Then HASHTAGS (0 or 3 from TAG_LIST).

TAG_LIST: [paste 3 to 5 tags you use]

Count the hook on desktop and on your phone after you paste. LinkedIn tests fold length. A hook that fits 210 characters on a laptop can truncate on mobile. If Instant writes a 280-character first sentence, send it back: "Cut HOOK to 200 characters. Keep the specific noun. Drop the moral."

Story posts need a time marker you own: "Last Tuesday" beats "recently" if the day is real. Lesson posts need one takeaway. Announcement posts need the product name, the date, and who it is for. Do not mix all three shapes in one draft. Pick one TASK.

Comments that add one useful point

A useful comment adds a fact, a caveat, or a question the original post did not cover. It does not repeat the author's thesis in new adjectives. Cap FORMAT at 400 characters unless the thread is a technical debate you can add to with a method you used. Ban "This." Ban "So important." Ban a mini-post that could stand as your own update.

Comment skeleton extra lines:

TASK: Write 2 to 4 sentences that add one point from STORY_FACTS to PARENT_POST. Do not summarize PARENT_POST. Do not pitch my product unless PARENT_POST asked for tools.

PARENT_POST: [paste the post you will comment on]

FORMAT: Max 400 characters. No hashtags. No emojis unless SAMPLE_A uses them. Open with the point. Skip greeting the author by name.

Paste the parent post. ChatGPT invents a reply to a phantom argument if you omit it. If you disagree, state the disagreement as a constraint: "I disagree with [claim]. Argue from STORY_FACTS. Stay civil. No dunking." Public disagreement without facts reads as a pile-on.

About-section drafts under the character cap

Treat About as a landing page. The first 230 characters must name who you help and the proof you own. The middle holds 2 to 4 short proof blocks from STORY_FACTS. The end holds how to reach you: email pattern, site, or "message me." Target 1,200 to 2,000 characters unless you have more proof that a reader would miss. Stay under 2,600.

About skeleton extra lines:

TASK: Draft my LinkedIn About from STORY_FACTS. Sentence 1 to 2 must stand alone as the mobile hook (max 230 characters). Then 3 proof paragraphs. Then a close with CONTACT. No resume dump. No third person.

CONTACT: [how people should reach you]

OUTPUT_SHAPE: HOOK / PROOF / HOW_I_WORK / CONTACT. Print a character count for HOOK and for the full draft.

Reuse STORY_FACTS for a headline in a follow-up message: "Write 5 headline options under 220 characters. Lead with [title or outcome]. No keyword stuffing. No vertical bars unless SAMPLE_A uses them." Check the first 60 to 70 characters, the span search results show. You type the final headline yourself from options you accept.

Step-by-step LinkedIn rewrite workflow

Use one chat thread per campaign or per About rewrite. Dumping a product launch, a hiring post, and a personal story into one thread muddies the reader. The loop below keeps STORY_FACTS stable while you swap TASK. You spend free ChatGPT or PromptMake runs on structure, then human time on fact checks and on reading the draft out loud. Skip the fact sheet and you will accept a metric you cannot defend in a comment thread.

Read the Instant draft in the LinkedIn composer, not in the chat. Line breaks change. The fold changes. A sentence that looked fine in a paragraph can look like a slogan once LinkedIn stacks it. Fix line breaks by hand. Send ChatGPT a screenshot-free note: "HOOK is wrapping at word 11 on mobile. Cut 20 characters. Keep the noun [X]."

Step 1: Build the fact sheet and voice sample

List proof in ugly notes. Example: "Acme, Product lead, 2023-2026. Shipped billing v2. Cut dunning emails from 4 to 2. Audience: SaaS ops managers." Mark uncertain numbers with a question mark. In the prompt, tell ChatGPT to keep those as [NEED FACT] or omit them. Guessing "3x pipeline" when you meant "more demos" creates a public correction you will hate.

Pick two posts you published that you like. Paste them as SAMPLE_A and SAMPLE_B. If you have no posts, paste a paragraph from an email you sent to a colleague. Spoken cadence beats a blank voice ask. Skip other people's viral posts as samples. ChatGPT will copy their hooks and you will sound like a clone of a stranger.

Step 2: Draft on GPT-5.5 Instant

Run the skeleton for one surface. Ask for 3 hook options if the surface is a feed post, then pick one and say "write the body under the chosen hook." Instant is fast enough for that fork. For comments, ask for 2 variants: one agreement-plus-add, one question. Pick one. Do not post both.

Optional scaffold: open https://promptmake.net/text, describe "LinkedIn [post | comment | About] with honesty rules, hook character cap, and voice sample," generate once, then paste STORY_FACTS into the returned structure. Guests get about three runs per day; free accounts get about five. Use a run to shape the prompt, then finish in ChatGPT.

Step 3: Audit voice and honesty on GPT-5.6

Take the winning draft into a second message, or switch to GPT-5.6 in the same thread:

TASK: Audit DRAFT against STORY_FACTS and SAMPLE_A. For each sentence, reply Keep, Soften, or Remove. Soften means the claim overreaches my facts or the cadence does not match SAMPLE_A. Propose a safer rewrite for Soften and Remove lines. Never add new achievements. Print hook character count again.

Read the audit. Type the facts you owe. Paste the composer text one last time and ask Instant for a line-break pass: "Keep words. Break after sentence 1, 2, and 4. No new claims." Then you post, comment, or paste into About. Save the winning prompt next to the surface name and model label so next week's posts reuse the wrapper.

Mistakes that wreck LinkedIn ChatGPT drafts

Mistake 1: Asking ChatGPT to "write a viral LinkedIn post" with no story. The model fills the void with a fake failure and a tidy moral. Paste a real incident or skip the post.

Mistake 2: Allowing invented metrics and client names. If you did not supply a number or a logo, ban both in FORMAT. Fake "saved a Fortune 500 40%" fails the first coworker comment.

Mistake 3: One prompt for LinkedIn plus Instagram plus X. Character limits and hook rules clash. Keep this workflow on LinkedIn. Use a multi-platform guide if you need other channels.

Mistake 4: Pasting a resume and asking for an About that "sounds executive." Title inflation is a public record. Match the title you hold.

Mistake 5: Using GPT-5.5 Instant as the last honesty check on a post that names a customer or a revenue number. Instant is fine for first drafts. Route the audit to GPT-5.6 when a claim could anger a client or a former employer.

Mistake 6: Commenting with a 1,200-character essay that restates the author's post and ends with your URL. Readers treat that as spam. Add one point. Leave the URL for your own update unless they asked.

Mistake 7: Trusting ChatGPT on dates, product names, and headcount. Type those from your records. Ask the model for verbs and line breaks.

Mistake 8: Hashtag walls and emoji stacks in the hook. They burn the 200-character budget and look like ads. Put tags at the end or omit them.

Mistake 9: Posting the chat output without reading it in the LinkedIn composer. Fold length and line breaks differ from the chat window.

GPT-5.5 Instant and GPT-5.6 notes for LinkedIn work (mid-2026)

ChatGPT defaults to GPT-5.5 Instant for fast chat. Instant fits brainstorming three hooks, first-pass posts, comment variants, and About bodies when STORY_FACTS and SAMPLE_A sit in the message. Keep prompts short: ROLE, TASK, FORMAT, FACTS, RULES. Skip long chain-of-thought slogans. Instant plus a few-shot sample of your line-break style beats a lecture on "thought leadership."

GPT-5.6 (Sol in API naming as of mid-2026) fits harder edit passes: honesty audits, contradiction checks between the draft and the fact sheet, and voice matching against SAMPLE_A without adding new wins. Give goal, constraints, and format. Drop "think step by step" padding on reasoning-class models. A customer named in a post, a revenue claim, or an About that a recruiter will screenshot is a GPT-5.6 audit job.

Terra and Luna appear in OpenAI's broader lineup; for consumer LinkedIn drafting you will live in Instant and the GPT-5.6 chat option more than API labels. Hedge on exact menu names in the ChatGPT UI. They shift. Re-check the model picker when you open a new thread.

Claude Sonnet 5 handles a long SAMPLE pack plus several parent posts if you batch comment drafts offline. Claude Opus 5 fits an About audit when the profile is client-facing. Gemini 3.5 Flash fits volume comment drafts from a list of parent posts you pasted. Each model needs your facts, and you are the one who posts.

Prompting split that holds: Instant and Flash get RTF plus SAMPLE_A when format matters. GPT-5.6, Opus 5, and Gemini 3.1 Pro get goal + constraints + format, with an explicit "never invent" rule and a [NEED FACT] token. Custom Instructions in ChatGPT can hold your banned-phrase list and "first person, short lines." Keep STORY_FACTS out of Custom Instructions. Facts go stale. Paste them per thread.

Build LinkedIn prompts with PromptMake

Write the rough ask in plain words: surface (post, comment, or About), hook cap, honesty rule, and whether you have a voice sample. Open https://promptmake.net/text and generate a structured prompt once. Expect labeled sections you can fill with STORY_FACTS, READER, and SAMPLE_A.

Edit product names, client names, and metrics yourself. PromptMake cannot know your numbers. Paste the filled prompt into ChatGPT on Instant for drafts or GPT-5.6 for audits. Keep free-tier runs for scaffolding. Skip five synonym retries of "make this go viral."

Workflow that sticks: fact sheet + voice sample → PromptMake scaffold → fill story and reader → Instant draft → GPT-5.6 honesty and voice audit → paste into the LinkedIn composer and fix the fold by hand. Store one template per surface so you do not rewrite ROLE and RULES from scratch each week.

FAQ

What are the best chatgpt prompts for LinkedIn in 2026?

The best chatgpt prompts for LinkedIn lead with ROLE and honesty rules, paste STORY_FACTS plus a voice sample, then demand FORMAT with a hook character cap for posts, a one-point rule for comments, or a 230-character About hook. Add a second audit prompt that marks Keep, Soften, or Remove against your facts. Match GPT-5.5 Instant for drafts and GPT-5.6 for the audit when you name a customer, a metric, or a title.

Can ChatGPT write my LinkedIn About section from scratch?

ChatGPT can draft structure and wording from notes you supply. It should not invent employers, dates, clients, or impact numbers. Start from a fact sheet you wrote offline. Treat a blank "write my About" prompt as high risk for fiction that a former colleague can contradict in one comment.

Should I use GPT-5.5 Instant or GPT-5.6 for LinkedIn prompts?

Use Instant for first hooks, post bodies, comment variants, and About drafts when STORY_FACTS and SAMPLE_A sit in the message. Use GPT-5.6 when you need an honesty audit, a voice match against your sample, or a contradiction check on a named customer. Run the same draft through both when the post is high stakes. Skip the second model when you are commenting with a two-sentence add-on that names no one.

How do I prompt ChatGPT for LinkedIn comments without sounding like a bot?

Paste the parent post. Ask for 2 to 4 sentences that add one fact or one question from your experience. Ban summaries of the author's thesis, ban "Great share," and cap length at 400 characters unless you have a method to add. Paste SAMPLE_A so the comment uses your cadence, then read it in the thread composer before you submit.

Can PromptMake help with chatgpt prompts for LinkedIn for free?

PromptMake /text turns a rough LinkedIn idea into a labeled prompt you can aim at ChatGPT. Guests get about three generations per day; registered free users get about five. Fill in your own STORY_FACTS and voice sample, then paste into Instant for drafts or GPT-5.6 for audits. Do not spend the free runs on "make it viral" retries.

Can I reuse one ChatGPT prompt for LinkedIn, Instagram, and X?

Keep this workflow on LinkedIn. Instagram captions and X posts use different fold lengths, hashtag norms, and tones. A single unstructured "all platforms" ask produces drafts that break at least one channel. If you need other networks, use a multi-platform template guide with a separate FORMAT block per channel, and do not paste a LinkedIn post into X and hope the hook survives.

How long should a LinkedIn post be if ChatGPT drafts it?

As of mid-2026, LinkedIn posts cap near 3,000 characters, and the feed hides most of the text after about 140 characters on mobile and about 210 on desktop. Put the hook inside that fold. Aim the body at 900 to 1,600 characters for a story or lesson post unless you have more proof the reader needs. Count the hook after you paste into the composer, and ask ChatGPT to reprint the count if you cut lines by hand.