How to export your LinkedIn connections

Getting the file

This has to be done on desktop — the mobile apps don't expose the data export at all. It takes about a minute of clicking and then a wait.

What's actually in it

Six columns, and that's the whole file:

• First Name and Last Name — exactly as the person typed them, suffixes, emoji and all
• URL — their public profile address, the most reliable way to identify someone later
• Email Address — blank for the large majority of people
• Company — their current employer, as free text
• Position — their current job title, as free text
• Connected On — the date the connection was made

That's it. There are no bios, no headlines, no post history, no mutual-connection counts and no company sizes. People often assume the export mirrors what they see on a profile page. It doesn't — it's the minimum LinkedIn considers yours to take with you.

Why the email column is mostly empty

LinkedIn only includes a connection's email address if that person has opted to share it with their connections. Most never change the default, so a typical export has addresses for somewhere between two and ten percent of rows. This is a privacy setting on their end, not a bug in your export, and there is no way to make it fuller.

Three quirks that break spreadsheets

1. There's a preamble before the header row

The file opens with a Notes: line and a paragraph of prose about missing email addresses, then a blank line, and only then the real column headers. Open it straight into Excel or Google Sheets and you get a single mangled column, because the importer treats that prose as your header row. Delete the first three lines before importing, or use a tool that looks for the First Name header itself.

2. The date format depends on your account locale

Most exports use 21 Nov 2015. US accounts emit 11/21/15. If you're sorting by connection date and the order looks random, this is why — your spreadsheet has parsed some rows as text and some as dates.

3. Job titles and companies are free text

There's no normalisation whatsoever. VP Ops, V.P. Operations, Vice President, Operations and VP of Ops (EMEA) are four different strings. The same is true of employers — Acme Ltd, ACME Limited and Acme will not group together in a pivot table. Any serious analysis has to normalise these before counting anything.

What you can legitimately do with it

This file is yours. Exporting it, storing it and analysing it are all squarely within LinkedIn's terms, because the export tool exists precisely so members can take their data with them.

What is not within the terms is going back to LinkedIn to enrich it — visiting each profile URL to scrape a headline or bio. That breaches the User Agreement, risks your account, and in 2025 LinkedIn sued and shut down Proxycurl, the best-known company doing it at scale. If someone offers to “enrich” your export with profile data, that data came from somewhere, and the somewhere is the problem.

The honest ceiling is this: title, employer, and how long ago you connected. That's less than people expect, and considerably more than most people ever use.

Cleaning it up before you analyse anything

Two normalisation jobs stand between the raw file and anything you can count. Both are boring and both are the difference between a useful list and a pivot table full of near-duplicates.

Normalising employers

Acme Ltd, ACME Limited, Acme Ltd. and Acme are one company and four strings. Lowercase everything, strip punctuation, then strip the legal suffixes — ltd, limited, inc, llc, gmbh, plc, bv, oy, ab, pty, corp, holdings, group. Accented characters need folding too, or Société Générale and Societe Generale stay apart. On a 3,000-row export this typically collapses 1,400 apparent employers down to about 1,000 real ones.

Normalising job titles

Harder, because there's no canonical list. In practice you don't need one — you need to match seniority and function well enough to filter. Search for the function words (ops, operations, revenue, finance) and the seniority words (head, director, vp, vice president, chief, founder, owner) separately, and treat the intersection as your candidate set.

What to do with the rows that have nothing

Every export has rows with a name and five empty columns. Don't delete them — they're real people you're connected to, and their absence from your analysis is a known gap rather than a mistake. But don't try to fill them either: the only way to do that is to go back to LinkedIn and read their profile, which is the line you shouldn't cross.