I Removed 21,000 LinkedIn Connections. Here's What Happened to My Reach.

I had 30,000 LinkedIn connections and thought I was doing everything right. More connections meant more reach, right?

Wrong. I ran a DeadWeight audit on my own network and discovered that 21,000 of my connections were dead weight. Ghosts. Inactive profiles. People who had not posted, liked, or commented on anything in over a year. Some had not been active in five years.

That is 70% of my entire network doing absolutely nothing for my content.

Why I Decided to Audit My Network

I noticed my LinkedIn impressions had been declining for months. Posts that used to get 5,000 impressions were barely hitting 1,500. I was posting the same quality content at the same frequency, but fewer people were seeing it.

The tipping point was learning how LinkedIn's algorithm actually works. When you publish a post, LinkedIn does not show it to all your connections at once. It tests it on a small sample of your network first. If that sample engages (likes, comments, shares), the algorithm expands distribution. If they do not engage, your post gets buried.

Here is the problem: if 70% of your connections are inactive, there is a very high chance your post gets tested on dead accounts. They will never engage. The algorithm sees zero interaction and decides your content is not worth showing to anyone else.

Your ghost connections are actively suppressing your reach.

What the Audit Revealed

I ran every one of my 30,000 connections through our scoring algorithm, which evaluates three dimensions:

Activity Recency

This is the biggest signal. We check when each connection last posted, liked, or commented on LinkedIn. The scoring:

No activity detected: +10 points (worst)
Last active 5+ years ago: +9 points
Last active 2-5 years ago: +7-8 points
Last active 6-12 months ago: +3-5 points
Active in the last 30 days: 0 points (best)

Of my 30,000 connections, over 14,000 had no detectable activity at all. Another 7,000 had not been active in over a year.

Profile Completeness

Incomplete profiles correlate strongly with inactive accounts. We check for:

No profile photo: +2 points
No experience listed: +1 point
No education listed: +1 point

A connection with no photo, no experience, and no education is almost certainly a ghost account that was created and abandoned.

Network Size

Connections with very small networks (under 100 connections) tend to be inactive or spam accounts. Large networks (500+) slightly reduce the score since these users are more likely to be active.

Tiny network (<100): +2 points
Large network (500+): -1 point

The Categories

Every connection gets a total score that maps to a category:

Ghost (7+): No activity, incomplete profile. Remove without hesitation.
Strong Remove (5-6): Very low activity. Almost certainly not seeing your content.
Likely Remove (3-4): Declining engagement. Worth reviewing but probably removable.
Review (1-2): Low but present activity. Check manually before deciding.
Keep (0 or below): Active, complete profile. These are your real network.

My Results

Out of 30,000 connections:

Ghost: 14,200 (47%)
Strong Remove: 4,100 (14%)
Likely Remove: 2,700 (9%)
Review: 3,800 (13%)
Keep: 5,200 (17%)

Only 17% of my network was genuinely active and worth keeping. The other 83% ranged from completely dead to barely alive.

The Removal Process

I did not remove all 21,000 at once. LinkedIn does not have a bulk remove feature, and removing too many connections too quickly can trigger their automated systems.

Here is the approach I used:

Start with Ghosts. These are the easiest decisions. No activity, incomplete profiles. I removed 200-300 per day.
Move to Strong Remove. Same process, slightly more judgment involved. Some had partial profiles but zero recent activity.
Review the Likely Remove category manually. Some of these turned out to be people I actually knew who just were not active on LinkedIn.
Keep the Review category for last. Many of these ended up being connections worth keeping after a closer look.

The whole process took about three weeks of 20-30 minutes per day.

What Happened to My Reach

The results were not instant. LinkedIn's algorithm takes time to recalibrate after significant network changes. But within two weeks of starting the removal process, I noticed clear changes:

Week 1-2: Impressions held steady despite having fewer connections. This alone was telling, since losing connections should theoretically reduce your audience.

Week 3-4: Impressions started climbing. Posts that had been getting 1,500 impressions were now hitting 3,000-4,000.

Month 2: Engagement rate (likes + comments relative to impressions) roughly doubled. The people seeing my posts were now people who actually use LinkedIn.

Month 3: Consistent 3-4x improvement in average impressions per post compared to before the cleanup.

The Counterintuitive Math

Going from 30,000 connections to 9,000 felt like a massive loss. But the math tells a different story.

Before: 30,000 connections, 1,500 avg impressions = 5% reach rate After: 9,000 connections, 4,500 avg impressions = 50% reach rate

My content was reaching 10x more of my actual network. And because those people were active users, they were engaging, which triggered even more distribution to second and third-degree connections.

A smaller, active network outperforms a large, dead one every time.

Lessons Learned

Accept every connection request is bad advice. The LinkedIn growth playbook of "connect with everyone" actively damages your reach over time. Be selective.

Your connection count is a vanity metric. Nobody cares that you have 30,000 connections if only 5,000 of them are real people who use the platform.

Regular audits matter. Connections go stale. Someone active today might abandon LinkedIn in six months. I plan to re-audit every quarter.

The algorithm rewards quality over quantity. This is true for content and connections. LinkedIn wants to show good content to engaged users. Help the algorithm by making sure your network is full of engaged users.

How to Audit Your Own Network

You can do a basic version manually by exporting your connections (Settings > Data Privacy > Get a copy of your data) and reviewing the CSV. But the export does not include activity data, profile completeness, or scoring. You are left guessing.

DeadWeight automates the entire process. You export your CSV, we enrich each connection with data from outside LinkedIn, and deliver an interactive dashboard where you can filter, sort, and review every connection before removing anyone.

The audit costs $197 and is delivered within 48 hours. If your network has more than a few hundred connections, the time savings alone are worth it compared to manually researching each one.

The Bottom Line

If your LinkedIn reach has been declining, your network is probably the problem. Most people have 60-80% dead weight in their connections. Removing them feels uncomfortable, but the algorithm rewards it.

The best time to audit your network was a year ago. The second best time is now.