How it works
Three steps. No gray areas.
ConnectionCleanup automates the finding, not the removing. Nothing it does ever touches LinkedIn's site, which is exactly why it can't get your account in trouble.
Import your data
LinkedIn gives every member a free export of their connections: name, company, title, profile link, and when you connected. After you sign up we show you exactly where to click, and email you a nudge 24 hours later in case the file lands while you're busy.
Drag the file into ConnectionCleanup and analysis runs instantly, entirely in your browser. The file is never uploaded to a server.
Drag & drop Connections.csv here
Parsed in your browser · never uploaded
Review what we found
Our recommended detection rules segment your network, and every flag carries a confidence level, so you always know what's safe to remove with minimal checking versus worth a profile glance first. Bulk-select by segment, keep or remove per person.
| Rule | What it looks for | Confidence |
|---|---|---|
| Title keyword match | The title contains a keyword you chose: “retired”, “student”, or your own list. | High |
| Your custom rules | Anyone whose title or company contains a keyword you set. “Remove all crypto founders” is three clicks. | High |
| Exact duplicates | Same name and same profile URL appearing twice in your export. | High |
| Retired / incomplete profiles | The title mentions retirement (any age), or the profile lists no title and no company at all. Profiles missing just one of the two get a medium flag instead. | High / Medium |
| Student-era titles | “Student”, “intern”, “graduate”, “trainee”, “apprentice”: high confidence when you connected 5+ years ago, medium at 3–5 years (long PhDs happen). | High / Medium |
| Job seekers who moved on | “Open to work”-style titles from 4+ years ago, and profiles still listing a university or school as their company after 5+ years. | Medium |
| Stale connections | Connected longer ago than a threshold you set (default: 5 years). | Medium |
| No change between exports | On a re-run with a newer export: title and company unchanged since last time, the only honest inactivity signal your data allows. | Medium |
| Possible duplicates | Same name, different profile URL. Could be two entries, could be two people. You decide. | Low |
What we don't claim: “inactive on LinkedIn”. The export has no activity data, so no tool can honestly detect that from it.
Illustrative example.
Clean up with the checklist
Every removal is your own click on LinkedIn's own button, in your own browser. The checklist just keeps you organized: who, why they were flagged, a link to their profile, and a progress bar that saves automatically.
Focus mode steps you through one profile at a time. Do 20 today and 100 next week; your place is kept. Export the final list as a CSV anytime.
Built-in pacing guardrails (on by default, off anytime) keep your browsing human: at most 5 profile tabs open at once, with a natural 10 to 15 second pause between opens.