Viveka 
Why is your LinkedIn connection count working against you?
Because LinkedIn’s ranking AI — the model it calls 360Brew — leans hard on engagement signals, and when a big share of your connections never interact with anything you post, that silence gets read as evidence your content isn’t worth showing to anyone. A large connection number used to be a vanity metric you could feel good about. Now it’s a reach problem: the dead-weight connections who never engage are actively holding your content back from the people who would.
One woman in the session had lived exactly this. She’d written about the problem months earlier, spent hours manually culling connections, then handed the job to a VA who promptly deleted the wrong people, and she was still stuck looking for something that actually worked. That’s the trap — the fix feels too tedious to do by hand and too risky to hand off blindly. This is the middle path.
How do you clean up your LinkedIn connections with Claude?
Three steps: download your connection data, curate the keep-or-remove list with Claude, then let the Claude Chrome extension do the removing at a human pace. None of it requires you to click through thousands of profiles one at a time.
- Download your archive. In LinkedIn, go to Settings → Data Privacy → Download Your Data, and request the full archive, not the mini version. It takes about 24 hours to arrive and will very likely land in your spam folder, so go look for it there.
- Curate the list with Claude. Upload the connections CSV into Claude and give it clear instructions about who to remove, who to keep, and what signals matter — alignment with your ideal client, engagement history, role type. Iterate on the criteria until the list looks right, and put your own eyes on it before you do anything, because this is exactly where the VA went wrong.
- Let the Chrome extension work. Open the Claude Chrome extension while you’re in LinkedIn, hand it the curated list, and let it remove connections in human time — deliberately not instant, which is what keeps you from tripping LinkedIn’s bot detection. Cap it at around a hundred at a time and keep an eye on it rather than walking away.
What criteria should you use to decide who to keep?
The two that matter most are alignment with the people you actually serve and engagement history. Ask whether a connection looks like someone in your ideal client world or a peer worth having, and whether they’ve ever engaged with your content; if the answer to both is no, they’re probably dead weight. Role type and mutual relevance help you break ties from there.
Whatever criteria you land on, review the final list yourself before a single person is removed. Claude can sort ten thousand connections in minutes, but you’re the one who knows that the quiet connection from 2014 is actually a referral source. The tool does the tedious part; the judgment stays yours. (If you want the fuller version of that principle, I made the case in using AI vs building with it.)
What’s the bonus hiding in your LinkedIn data archive?
The same archive download includes every article you’ve ever published on LinkedIn, and if you were writing there for years before AI existed, that is a goldmine of authentic voice data. Pull those articles out, feed them to the AI tool you use, and let it study how you actually write — your rhythm, your word choices, the way you open and close.
Years of pre-AI writing is far more valuable for voice calibration than most people realize, precisely because no machine had a hand in it. It’s the cleanest sample of your real voice you’ll ever have, and it’s sitting in a file you already have the right to download. That’s the same instinct behind telling the stories only you could tell — the raw material is already yours.
What should you do this week?
Start the clock on the slow part. Request your LinkedIn data archive today so it’s ready when you are, and while you wait, write your keep-or-remove criteria before you ever touch the list — alignment and engagement are the two that carry the most weight. When the archive arrives, test the Claude Chrome extension on a small batch first and watch it work before you trust it with more, and pull your article archive while you’re in there so you’ve got your voice data set aside.
One firm caution: don’t point this at DMs or outreach. Cleaning up your own connection list is a housekeeping task LinkedIn has no problem with; automating messages to other people is a different thing entirely, and LinkedIn will not thank you for it.