Claude Prompt Templates for LinkedIn Posts

Why does AI-written LinkedIn content so often sound the same regardless of who's posting it? Scroll through a LinkedIn feed and you'll spot the tells within seconds: one-line paragraphs stacked like a ladder, a hook that starts with a number or a question, a bullet list in the middle, and "Agree? Drop a comment below." at the end. This isn't Claude's fault. It's a prompt problem -- most LinkedIn prompts don't specify what makes this person's voice different from the average LinkedIn voice, so you get the average LinkedIn voice back.

The Problem a Consultant Faced

An independent strategy consultant had been using AI to help draft LinkedIn posts for three months. Her engagement was mediocre. When she looked honestly at the posts, they sounded like they could have been written by anyone with a management consulting background -- which meant they were competing with every other management consultant on LinkedIn instead of standing out as her specific perspective.

Her posts had ideas but no voice. They had structure but no texture. They were, in the language of the LinkedIn algorithm, getting low dwell time because readers scanned, saw nothing they hadn't seen before, and kept scrolling.

The Wrong Approach

Write a LinkedIn post about [topic].

This produces a LinkedIn post. It will have a hook. It will have some content. It will have a call to action. It will be indistinguishable from the posts on either side of it in someone's feed. Technically a post. Actually invisible.

The Correct Prompt

Write a LinkedIn post for a [your role/expertise] with this specific point of view: [your actual opinion on the topic, including who you disagree with or what conventional wisdom you're pushing back on]

My voice (write like this, not like generic LinkedIn content):
[paste 3-5 sentences from a LinkedIn post or message you've written that felt most like you]

Hook style for this post: [pick one -- contrarian opener, specific story, surprising fact, direct take]
Target reader: [specific -- e.g., "VP of Sales who's frustrated that their team isn't adopting a new CRM"]

Structure:
- Hook: 1-2 lines that earn the scroll-stop
- Build: 3-4 short paragraphs, each advancing the argument (not just listing points)
- Close: a specific concrete takeaway, not a question fishing for engagement

Hard rules: no numbered lists, no "Here's what I learned:", no "Agree?" at the end, no emojis

What this does: The hard rules list actively fights against the AI tells that make LinkedIn content look AI-generated -- each of those banned elements is a pattern the algorithm has seen enough times that it's trained readers to skip them. Banning them forces the output toward patterns that actually get read.

⚡ Pro tip: The voice sample (3-5 sentences from something you actually wrote) is worth more than a detailed description of your voice. "Conversational but direct" tells Claude almost nothing. Actual sentences you wrote tell it your sentence rhythm, your vocabulary range, your level of technical specificity, and your relationship to hedging language -- all of which are impossible to describe but easy to pattern-match from examples.

Results and What Changed

After switching to this prompt structure, the consultant's posts got materially longer dwell times. More importantly, she started getting inbound messages from people who'd seen a specific post and wanted to discuss the specific point she'd made -- a signal that the content was actually being read, not just scrolled past. The content hadn't become more impressive. It had become more recognizably hers, which meant it connected with people who shared her specific perspective rather than everyone in general and no one in particular.

How to Apply This to Your Situation

For repurposing existing content into LinkedIn posts:

Here's [an article / interview transcript / talk I gave]: [paste or summarize]

Extract the single most counterintuitive or surprising point from this for a LinkedIn audience of [describe your typical followers]. Don't summarize the piece -- find the one thing that would make someone stop scrolling.

Write a LinkedIn post built around that one point only. If the piece doesn't have a genuinely surprising or counterintuitive point, tell me that honestly -- a summary post with no real hook is worse than not posting.

What this does: The instruction to flag if there's no genuinely surprising point is an honest quality filter -- it prevents posts that are content for the sake of posting rather than content worth posting, which is the actual cause of declining engagement for most LinkedIn accounts over time.

⚠️ Common mistake: Using the same hook style for every post. Your regular readers notice this after 3-4 posts. Rotate deliberately between story, contrarian take, surprising stat, and direct question -- not randomly, but based on which hook style fits the specific content best.

A Performance Diagnosis Prompt

Most LinkedIn creators have a post that performed unusually well and one that flopped -- but rarely investigate why, beyond a vague sense that "that one resonated." A prompt-based diagnosis extracts the actual lesson:

Here are two LinkedIn posts I've written. One performed well [describe: shares, comments, DMs], one didn't [describe performance].

Good post: [paste]
Poor post: [paste]

Analyze what's structurally different between them -- not just what each says, but how they're constructed: hook approach, sentence structure, level of specificity, whether the argument is self-contained or requires context, whether they end with a question or a statement.

What's the most likely structural explanation for the performance gap, based on what you can observe? Give me a hypothesis, not a certainty.

What this does: Getting a structural diagnosis (hook approach, sentence rhythm, self-containment) rather than a content one ("the good post had a better idea") is more actionable -- you can replicate structure deliberately, but you can't replicate having a better idea on command.

⚡ Pro tip: Run this comparison diagnosis once a month across your last 8-10 posts. The pattern that emerges over 3-4 months is a better guide to what to keep doing than any single post's performance -- one viral post is an event; a consistent pattern is a system.

Next Steps

A claude prompt template for LinkedIn posts is worth maintaining as a living document: when a post performs unusually well, note what made it different (hook style? level of specificity? emotional register?) and encode that as a refinement. When a post underperforms, diagnose whether it was the content or the framing. Over time, this turns a generic template into a genuinely calibrated system for your specific voice and audience -- worth keeping in PromptABCD where you can refine and access it consistently.

The performance diagnosis prompt is genuinely more useful run quarterly across a full set of posts than run on individual posts in isolation -- patterns that aren't visible post by post become clear when you look at 20 posts together. Hook styles that consistently outperform regardless of topic, levels of specificity that correlate with engagement, whether story-based posts outperform opinion-based posts for your audience -- these are the findings that turn a posting strategy from intuition into something you can actually replicate deliberately. Most LinkedIn creators who post consistently have this information available in their analytics but never look at it systematically. The diagnosis prompt makes that analysis a 10-minute quarterly exercise rather than a manual data project.

One more practice worth building in: for every post that generates an unsolicited direct message from someone who found it genuinely useful, note the topic, angle, and hook style in your template notes. DMs are a higher-quality signal than likes or comments because they require more friction from the reader -- someone who takes the time to message you directly found the content specifically valuable to their situation, not just interesting in passing. Enough of those signals over time tells you exactly what your audience most wants you to write about next.

And actually, the relationship between posting consistency and template quality runs in both directions. The template makes consistency easier by removing the decision fatigue of starting from scratch each time. But consistency is also what gives you the data to improve the template -- without a steady stream of posts, you don't have enough performance variation to diagnose what's working and why. The two practices compound together in a way that neither does alone. Build the template first; let the data improve it over time.