Between 2015 and 2019, LinkedIn ran a series of randomized experiments on the algorithm that decides who it suggests you connect with. Researchers from LinkedIn, Harvard Business School, Stanford and MIT analyzed the randomization those experiments created and published the result in Science in September 2022. The experiments varied the prevalence of weak ties in the networks of more than 20 million people over a five-year period, during which 2 billion new ties and 600,000 new jobs were created (science.org).
One sentence in it should change how you spend Tuesday mornings. Measuring the strength of a connection by how many contacts two people already had in common, the experiments found the benefit rises and then turns over: “adding ties with more than ten friends in common reduced the probability of a job transmission” (science.org).
More overlap, worse result. Past roughly ten shared names, a new connection stops carrying anything your existing circle did not already have, and the curve bends the wrong way.
What the experiment measured, and what it did not
Be precise about this before building anything on it, because the number is easy to stretch. The outcome the researchers counted was a job transmission: person A reports working at a company, person B later reports working at the same company at least a year afterward, and the two were already connected a full year before that (science.org). That is people moving into jobs on a professional network. It is not customers being referred to a plumbing company.
So the finding does not promise that a referral partner with eleven mutual contacts will send you fewer jobs. What it establishes is narrower and more useful than a promise. When the value of a relationship is the information it carries, overlap is the thing that destroys that value, and overlap is measurable before you spend a single hour on the relationship.
The size of the effect belongs to LinkedIn’s data. The counting is yours, and it is free.
The line almost nobody quotes, and it may reverse the advice for you
Weak-tie advice gets repeated as though it applies to everybody in every trade. The paper says plainly that it does not. From its own abstract: “Whereas weak ties increased job mobility in more digital industries, strong ties increased job mobility in less digital industries” (science.org).
Read that twice if you run a roofing company, a dental practice, a landscaping crew, a restaurant or a repair shop. In the less digital half of the economy, the experiments found the opposite of the advice that gets quoted at you. The ties that moved people into work there were the strong ones.
That single clause splits every reader into two groups running two different plays. Nearly all the networking advice in circulation is written for one of them and handed to both.
Which play you are running
If most of your work is found, sold and delivered through a screen, you are in the more digital half. Consultants, agencies, software, e-commerce, bookkeepers, remote services. Your constraint is novelty. The room you have attended every Thursday for two years has stopped producing information, because everyone in it already knows everyone you know. Refresh it.
If your work is local, physical and handed over in person, you are in the less digital half. Trades, food, clinics, salons, retail, in-home services. Your constraint is trust, and a strong tie is what trust looks like on a network diagram. Deepen a smaller number of relationships and stop buying volume.
Getting this backward is expensive in a specific way. The digital business that deepens ends up with four friends and no new information. The local business that refreshes ends up with two hundred contacts and nobody who will vouch for it.
Guidance and doors are not the same network
There is a second sorting to do, and it cuts across the first one. The prompt that this article grew out of states the goal in a single line: develop relationships with mentors, peers, and advisors who can provide guidance and open doors. That is two jobs, and they belong to two different sets of people.
Guidance is a small, strong, slow network. You want people who know your numbers, who will say the unwelcome thing, and who are still around in three years. Doors are close to the opposite. That network is wide, deliberately shallow, and its whole value is that those people do not know each other and do not already know what you know.
Most owners try to extract both from the same weekly group, then wonder why the guidance stays polite and the doors have all been walked through twice. One room cannot be both, because the property that makes a guidance relationship good is exactly the property the experiment found reduces the door-opening effect.
The guidance half is already paid for
The guidance network is the one owners most often skip, usually on the assumption that real advice costs money. In the United States, a large part of it does not.
SCORE, one of the Small Business Administration’s resource partners, states it directly: “SCORE mentoring is a free service offered to any aspiring or established owners of a U.S.-based business”, delivered one to one, with virtual or in-person options, and the same page adds that “Mentoring is always free, for the life of your business” (score.org). Nonprofits are named on that page as eligible alongside businesses.
SCORE is one of four resource partners. The SBA’s local assistance page lists Small Business Development Centers, SCORE, Veterans Business Outreach Centers and Women’s Business Centers, and says that “SBA and our network of partners offer free or low-cost counseling and training in your area” (sba.gov). The four differ mostly in who they are built for and how they deliver. Mentoring is ongoing and one to one, while development centers lean toward advising plus classes.
None of that is a substitute for a peer who runs a business your size in your town. It is, however, a real advisor with no invoice attached, and it takes a form and a zip code to start. If the guidance you need is aimed at somebody on your payroll rather than at you, the same free sources do that work too, which is the spine of our piece on writing a professional development plan as a small business playbook. If you run a nonprofit, the advisors question has a paperwork answer as well, and we walked it through in nonprofit board responsibilities and your own Form 990.
Four moves
1. Count the overlap before you spend the hour. On LinkedIn, a person’s profile shows how many connections the two of you already share, so the number is on the screen before you send anything. Off LinkedIn the count is manual and takes a minute: name the people you both know. Under about ten, this person sits outside your circle and can carry something into it. Well over ten, they are already inside it, and the coffee is a social event rather than a business one. Both are fine. Only one of them is work.
2. Book the free advisor before you buy the paid room. Request a SCORE mentor at score.org and a session at your nearest Small Business Development Center through the SBA’s local assistance page, and do it before you renew a membership or buy a table at anything. Two free sessions will tell you what kind of guidance you actually need, which is the question a paid room will never answer honestly.
3. Make one ask a month, and make it answerable in one sentence. “Let me know if you hear of anything” cannot be acted on, so nobody acts on it. “Do you know anyone who manages the maintenance contracts for a commercial building here in town?” can be answered yes or no in three seconds, and a yes has a name attached. One specific ask a month, to one person who is outside your circle, is a heavier program than it sounds. It is also the version of acquisition that compounds rather than resetting every month, which is the argument in our piece on how to get customers without starting over each month.
4. Put a date on refresh or depth, and pick only one. Once a quarter, open your calendar. If you are in the digital half, the question is how many of last quarter’s conversations were with people whose mutual-contact count was under ten, and the honest answer is usually zero. If you are in the local half, the question is which three relationships you moved from acquaintance to genuinely strong, and the answer is usually also zero. Write the number down. Next quarter it can be one.
Where AI fits, and where it does not
The useful AI job here is preparation, not contact. Before a meeting you can ask a general assistant to turn everything you know about a business into three specific questions worth asking its owner, or to compress your own year into two sentences a stranger could repeat accurately to someone else. Both are drafting tasks, both are checkable by you in under a minute, and both fix the real failure in most networking conversations, which is that the owner has not decided in advance what the conversation is for.
What AI should not do here is send the message. The one property that makes a weak tie valuable, according to the experiment, is that it reaches somebody your existing circle does not reach. Bulk-sending the same note to four hundred people does not create that property. It creates four hundred copies of a note.
What to track it in
A short list with dates beside it does more than any of these tools. If you want software anyway, the honest range is wide and starts at zero. The prices below were read off each vendor’s live pricing page in September 2026.
Google Contacts comes with any Google account at no cost and holds a note and a date per person, which is the entire requirement. If you want the business version with your own domain, Google Workspace Business Starter is $7 per user per month, and Business Standard is $14, both currently discounted for the first three months (workspace.google.com).
HubSpot gives its free tools to up to 2 users at $0 per month, which covers an owner and one other person. The paid Starter Customer Platform is listed at $20 per month per seat, currently promoted at $7 (hubspot.com).
Notion is free for one person, and Plus is $10 per member per month, with the page noting savings of up to 20 percent for paying yearly (notion.com).
The upsell to resist is the one that sells you sequences and automated follow-up. Twenty names you actually call do not need a pipeline. Two thousand names in a pipeline are not a network, and the experiment above is a reasonable argument that most of them are carrying nothing.
A disclosure about the prompt page
The prompt behind this piece, Build Your Strategic Network, is a five-step chain that walks an AI assistant through the principle and out to a plan, and it is worth a disclosure: its headline, its description and its prompt text are all about building a network of mentors and advisors, but the three “About This Prompt” panels on that page currently describe a completely different subject, product and customer profitability analysis. We have now checked twenty-one prompt pages in this family and eighteen carry a panel mismatch of some kind, so read the prompt itself rather than the panels. The wider library sits at businessprompter.com.
Our read, and it is a read
Facts above, judgment here.
Almost every piece of networking advice in print is written for somebody looking for a job, including the research this article is built on. Owners are not looking for jobs, and the advice arrives at them unedited. That is why it lands as “meet more people”, which is the one instruction the experiment does not support in a business the size of most of ours.
The part that does transfer is the arithmetic, and it points somewhere unfashionable. A network is not a sales channel with a friendlier name, and treating it as one is what makes owners quietly hate networking. It is capacity. A second person who can vouch for you, a second person who knows how to price the awkward job, a second person who covers the emergency: those are the things that let a small team accept work it currently turns down, and they arrive through people rather than through software. The right measure of a network is not leads generated. It is the number of things you can now say yes to.
Which is also the argument against the version being sold hardest right now, where a tool sends the messages, books the meetings and keeps the relationship warm on your behalf. Delegate the preparation. Keep the conversation. The conversation is the entire asset.
Frequently Asked Questions
How many mutual connections is too many?
In LinkedIn’s experiments, adding ties with more than ten friends in common reduced the probability of a job transmission (science.org), so ten is a reasonable working line rather than a law of nature. It was measured on job moves within one professional network, not on customer referrals. Use it as a filter for where to spend a scarce hour, not as a rule about who is worth knowing.
Does weak-tie advice apply to a local trades business?
Probably not in the form it usually arrives. The same study found that weak ties increased job mobility in more digital industries while strong ties increased job mobility in less digital industries (science.org). If your work is local and physical, the evidence points toward deepening a small number of relationships rather than collecting a large number of shallow ones.
Where can a small business owner get free advice from an experienced advisor?
SCORE offers free one to one mentoring to any aspiring or established owner of a U.S.-based business, virtually or in person, and states that mentoring stays free for the life of the business (score.org). The SBA also lists Small Business Development Centers, Veterans Business Outreach Centers and Women’s Business Centers as partners offering free or low-cost counseling and training locally (sba.gov). Start with a zip code on either site.
What should I actually ask for at a networking event?
Ask something that can be answered yes or no in three seconds and that produces a name when the answer is yes. A question naming a specific role, a specific type of company and a specific area gets acted on. A general request to be kept in mind does not, because there is nothing for the other person to do with it.
The two plays in this article are easy to tell apart on paper and uncomfortable in practice, because most of us are somewhere in between. So a genuine question for owners reading this, and the comments are the right place for it: when you last got real work through somebody you knew, was that person deep in your circle or barely in it at all? The answer decides which half of this article was written for you.
