LinkedIn keeps a record of just about every action you have ever taken on the platform. Every post, every comment, every message, every connection. Most people never touch it. That is a mistake, because that archive is one of the richest first-party datasets you own, and AI tools like Claude can turn it into real pipeline leverage in an afternoon.
Here are three ways I am using my own LinkedIn export right now, including the exact prompts, so you can run the same plays yourself.
How to export your LinkedIn data
Before anything else, request your archive. Go to Settings & Privacy > Data privacy > Get a copy of your data and select the full archive. LinkedIn emails you a download link, usually within 24 hours. Unzip it and you will find a folder of CSVs: posts, comments, messages, connections, reactions, and more.
Three files matter for what follows: Shares.csv (your posts), Comments.csv, messages.csv, and Connections.csv. Upload the relevant ones to Claude alongside each prompt below.
Build a tone of voice document
If you use AI to draft content, or you have anyone ghostwriting for you, the biggest problem is sounding like everyone else. The fix is a tone of voice document built from your actual writing, not from adjectives you picked off a list. Your LinkedIn archive contains years of it: posts written for reach, comments written in the moment, and DMs written one-to-one.
Upload your posts, comments, and messages files, then run this:
Prompt:
"I have uploaded CSV exports of every post, comment, and message I have written on LinkedIn. Analyze my writing style across all three files and produce a tone of voice reference document I can reuse when drafting content. Cover: typical sentence length and structure, vocabulary and phrases I repeat, how I open and close posts, formatting habits like line breaks and lists, where I use humor versus directness, and how my tone shifts between public posts, comments, and private messages. End with a set of do and don't rules, plus five example opening lines written in my voice."
Save the output as a standing reference. Every future draft, whether written by AI, a teammate, or an agency, gets checked against it. The result is content that ships faster and still sounds like you.
Find what content actually performs
Most people decide what to post based on gut feel and whatever did well last week. Your archive lets you replace gut feel with evidence across your entire posting history: timing, format, hooks, and topics, all matched against real engagement.
Upload your posts export and run this:
Prompt: "Attached is my LinkedIn posts export with dates, post content, and engagement data. Analyze what drives performance across my full history. Break down: engagement by day of week and time of day, performance by post format and length, the hook styles used in my top ten posts versus my bottom ten, and topics that consistently overperform or underperform. Then give me a ranked action list: what to double down on, what to kill, and what to test next, with the data behind each recommendation."
The goal is a feedback loop. You are not looking for one viral trick. You are looking for the three or four repeatable patterns that reliably earn attention from the people you want in your pipeline, so you can accelerate what works and stop spending effort on what does not.
Segment your network for outbound
This is the play with the most direct pipeline impact. Your connections file is a list of people who already accepted a relationship with you. Cross-reference it against your message history and you will find something valuable: connections who fit your ICP that you have never once messaged. Warm-ish, qualified, and completely untouched.
Upload your connections and messages exports and run this:
Prompt: "Here are my LinkedIn connections export and my messages export. Cross-reference them to identify every connection I have never exchanged a message with. Then segment that list three ways: by ICP fit, by geography, and by job title. My ICP is [describe your ideal customer: industry, company size, roles, region]. Output a prioritized table of never-messaged connections who fit my ICP, ranked strongest to weakest, with a short column explaining why each person fits."
Take the top tier and move them into a dedicated sequence, separate from your main inbox. That keeps active conversations clean while a structured cadence works through the people you should have talked to years ago.
Why this matters for pipeline
None of this is a novelty exercise. A tone of voice document makes every piece of content you produce more consistent and faster to ship. Performance analysis stops you wasting effort on posts that look busy but move nothing. Network segmentation turns a dormant asset into a live outbound list.
The bigger point: the data was always there. The barrier was never access, it was the hours of manual analysis nobody was going to do. AI removed that barrier. The people who win now are the ones who actually go get the file.
If you want this thinking applied to your whole go-to-market, not just your personal profile, that is the work we do at Marketing Copilot. We identify your tier-one accounts, find the right people inside them, and build the systems that reach only those people. If that sounds like a better use of your data, let's talk.
.png)
.png)
