There's an audit we run early with most clients that nobody asks for. We open their connections list, sort by most recent, and read the first fifty names out loud.
It is almost never a list of buyers.
It's freight brokers who connected to pitch. Software SDRs. Two students. A lead-gen agency that connects at scale and messages within four minutes. Someone's VA. Three other founders in an adjacent category. A recruiter for a role the founder has never held.
The founder's reaction is always the same, and it's always some version of: yeah, I just accept everything.
That's the finding. Not that the network is bad — that nobody ever decided what it should be. It was assembled by reflex, one click at a time, over four years, in the two seconds between opening the notifications tab and closing it.
Your first-degree network is a standing launch pool, not a contact list
Founders file connections under "networking." The platform files them under distribution.
Two mechanics matter here, and neither is controversial:
Every connection is automatically a follower. Connections are bidirectional. Followers are one-way. You can have followers who aren't connections, but you cannot have a connection who isn't a follower. So your first-degree network is a floor on who is eligible to see your work.
Posts don't reach everyone at once. They go to a subset first, and what that subset does with the post shapes what happens next. People who have a relationship with you are disproportionately in that early group — that's what a relationship signal is for.
Put those together and your connections list stops being a rolodex. It's the room your work walks into first, every time you publish.
We're not going to hand you a percentage for how heavily first-degree weighting counts. Nobody has published one, and anyone quoting you a specific number invented it or repeated someone who did. We've stopped repeating figures we can't produce a source for.
What needs no study is the part that should bother you: the composition of the room that sees you first was set by a reflex, not a decision. And you can go look at it in ten minutes.
Ecommerce founders accumulate a very specific kind of junk
Every professional on LinkedIn gets connection spam. Ecommerce founders get a particular blend of it, because the platform is where their entire vendor ecosystem prospects.
The recurring composition we see:
- 3PL, freight forwarding and customs brokerage reps. High volume, and they connect to every founder whose title mentions a product category.
- Software SDRs. Repricers, inventory tools, review tools, PPC platforms, listing optimizers. Some of these are people you should know. Most connected to run a sequence.
- Agencies pitching the service you already have in-house, or the one you sell.
- Job seekers and students, often in bulk after a hiring post.
- Other people's growth automations, which are not people at all.
None of these are shameful. Some become genuine relationships — we've hired people who first arrived as a cold connection request, and so have most of our clients. The problem isn't that these connections exist. It's the ratio, and the fact that the ratio was never chosen.
The practical consequence is that a meaningful share of the room reading your post first has no interest in the topic and no ability to buy. They didn't connect to read your work. They connected to send you a message. When they scroll past, the post gets graded on their indifference, and it's an indifference you recruited.
The obvious fix got materially more expensive in 2024
So prune it. Obvious answer, and it used to be a cleaner trade than it is now.
The widely reported behaviour — confirmed across LinkedIn tooling documentation and consistent with what we see in client accounts — is that around late September 2024 the removal mechanic changed. Previously you could remove a connection and the person could remain a follower. Now, removing a connection also removes the follow.
Verify this against your own account before you act on it, because platform behaviour changes without announcements and this one was never a press release. But the direction is not in dispute, and it reprices the decision entirely.
Before: pruning was almost free. You dropped the direct relationship and kept the audience.
After: every removal costs you a follower. You are not tidying a contact list, you are shrinking your distribution.
This is the part founders get wrong in both directions. Some are still working off the old mechanic and prune aggressively, then wonder why their follower count went backwards in a month where they published well. Others hear "removing costs you reach" and conclude the whole network is untouchable.
Neither is right. It means the removal button is now a real cost with a real benefit, which is to say it's a decision, and it needs a threshold.
Unfollow is the free lever. Remove is the expensive one.
These get used interchangeably in conversation and they do completely different jobs.
Unfollow changes what you see. Their posts leave your feed. You stay connected. They still see your posts. They are not notified. This costs you nothing.
Remove changes what they see. The connection ends, and on current behaviour the follow ends with it. They aren't notified either, but you've spent something.
So the sequencing is straightforward:
Unfollow liberally. This is the input-side fix and it's genuinely free. If your feed is 90% peers talking about the industry rather than customers talking about problems, that's not a content problem, it's a diet problem — and it leaks into your writing within about a month. You start referencing industry discourse instead of buyer situations, because industry discourse is what you've been reading. Unfollow is the cheapest correction available on the platform and almost nobody uses it, because it feels like a judgement about a person when it's a decision about your own attention.
Remove sparingly, and only for a reason you'd say out loud. Our threshold for clients is short: automated accounts, anyone who has used the connection to send something abusive or dishonest, and direct competitors who are clearly monitoring rather than reading. That's it. "Not my buyer" is not a removal reason — it's a reason not to have accepted, which is a different button.
Do not run a mass prune. We have never seen one pay for itself. A founder spends a Sunday removing four hundred connections, loses four hundred followers, and gets back a marginally cleaner room that nobody outside the account can perceive. The material improvement comes from the next thousand accepts, not the last four hundred.
The acceptance standard
Since the undo got expensive, the leverage moved to the front of the process. This is the whole fix, and it takes about six seconds per request.
Write down what you're accepting for. Three categories cover most founders:
- Buyers and buyer-adjacent. Anyone in a company that could hire you, buy from you, or refer someone who would. Accept.
- Genuine industry. Peers, partners, suppliers you actually work with, people whose work you'd read. Accept, and know that this group drives engagement and distribution more than pipeline — that's fine, it just isn't the same job.
- Everyone else. Ignore, or follow-back without connecting if you want to keep the door open.
The one rule we'd add for ecommerce founders specifically: if the request comes with a note that is a pitch, it isn't a connection request. It's a cold email that has asked you to permanently add the sender to the room where your work debuts first. Ignoring it costs you nothing and doesn't close anything — they can still follow you, message you through other routes, and connect later when there's an actual reason.
Then batch it. Once a week, four minutes, in the connections tab rather than in the notification bar between meetings. The failure mode isn't that founders have bad judgement about who to accept. It's that the accept button lives inside the same interface where they're clearing notifications, and clearing notifications is a task whose entire objective is to make things go away.
What this doesn't fix
Be honest with yourself about the size of this lever.
It won't rescue a posting problem. If you're publishing twice a month, the composition of your first-degree network is not your constraint. Cadence is. Fix the bigger thing first.
It won't move numbers this quarter. This is a compounding hygiene decision, not a growth tactic. The founders who benefit are the ones who ran a standard for a year, not a weekend.
It won't make peers into buyers. Your engagement will still skew toward people who look like you, because commenting is a public act that earns them something and buyers usually have a reason to stay quiet. The accept standard doesn't change who engages. It changes the floor on who's eligible to see you at all.
FAQ
Should I remove connections who never engage? No. Silence is not a signal about value — a meaningful share of the people who eventually hire you will never react to a single post. Removing quiet connections is the most reliable way to delete your buyers and keep your commenters.
Does a big network with low engagement hurt my reach? You'll see confident claims that it does, with specific percentages attached. We haven't found a source for any of them and we'd treat them the way we treat every unsourced platform number. What we'd say instead: a room stacked with people who didn't come for the work produces a weaker early signal than one that did, and that's true whether or not there's a penalty attached.
Should I accept competitors? Usually yes, and stop worrying about it. They can read your public posts either way. The one exception is if a competitor is using the connection for access to something non-public — messaging your clients, scraping your engagers. That's a removal reason.
I've got 8,000 connections built with no standard. Where do I start? Not with the archive. Start with the accept standard from this week forward and unfollow anything cluttering your feed. Then leave the back catalogue alone. Ninety days of deliberate accepts outperform a weekend of deletions, and they don't cost you a follower to do it.
Most founders treat their connections list as a record of who they've met. It's closer to a standing invitation list for everything they'll publish next year — and on current mechanics, it's a lot easier to add a name than to take one back.
If you're building an audience that's supposed to produce pipeline rather than applause, that distinction is worth six seconds a request.
We build and run LinkedIn content systems for ecommerce founders and Amazon operators — including the unglamorous parts. If you want a second opinion on what your audience is actually made of, get in touch.