The interesting question is narrower. Which relationships, maintained how, produce what. And on that the best available research says something most people find counterintuitive, which is that the relationships that generate opportunity are mostly the ones you are not close to.
The strongest evidence points away from intuition
In 2022 a team led by Karthik Rajkumar and Guillaume Saint-Jacques, with Sinan Aral as senior author, published a causal test of this in Science, using more than 20 million LinkedIn users tracked over five years. LinkedIn's connection recommendation system varied how many weak tie and strong tie suggestions users saw, which created a natural experiment, and the researchers could then separate the effect of how many new connections somebody made from the effect of how close those connections were.
Weak ties caused more job mobility than strong ties. Not correlated with. Caused, as well as a study of this kind can establish causation.
The shape of the finding is the useful part. The most valuable connections were moderately weak ones, where two people shared roughly ten acquaintances. Very close and very distant both did less. And when the researchers measured closeness by how often people actually interacted, the most useful contacts were the ones contacted least often.
I want to be straight about what this is and is not. It measures job mobility among people on one professional platform, not customers hiring a local business, and I am using it as an analogy rather than as direct evidence about your referral pipeline. The researchers themselves note it studied successes rather than the failures preceding them, that it was a natural experiment rather than a true randomized one, and that the effects were larger in digital industries than in physical ones. Other researchers have pointed out that people relying on tight knit communities may have less access to the wide shallow networks the study rewards.
Take it as the best thing we have on the mechanism rather than as a law.
Why it works that way
The mechanism is almost boringly simple once stated, and it is about information overlap rather than affection.
The people closest to you know what you know. They talk to the same people, hear the same news, and are aware of the same opportunities, because closeness in a network means shared surroundings. When your best friend learns that somebody needs a website built, you probably learned it too.
Somebody at the edge of your network occupies a different information neighborhood. They hear things you will never hear. When they learn that somebody needs a website, that information reaches you only through them, and there is no second path.
So the value of a relationship for generating opportunity is not how much the person likes you. It is how much they know that you do not. Those two things are close to inversely related, which is why the advice to deepen your closest relationships, while good advice for a life, is weak advice for a pipeline.
What this adds to what is already known
Word of mouth dominating small business acquisition is well established, and I have written about where small business customers actually come from and the gap between customers who say they would refer you and customers who actually do.
That work is about activation. People are willing and not prompted, and asking closes the distance.
This is a different point about structure. Even with perfect activation, your referral capacity is bounded by the shape of your network, and most owners have a network shaped badly for it. Deep in the middle, thin at the edges, because the middle is where the pleasant conversations are. You can ask your twelve closest contacts perfectly and still be asking twelve people who all know each other.
The part that gets left out
Relationship driven businesses have two costs that nobody mentions while recommending them.
The first is a ceiling. There is a limit to how many relationships one person can genuinely maintain, and every hour spent maintaining them is an hour not spent delivering. A business acquiring customers purely through personal relationship is capped by the owner's social capacity, and that cap arrives sooner than people expect. The businesses that grow past it do so by building something that generates demand without the owner in the room.
The second is concentration. If four relationships produce most of your revenue, you do not have a diversified business, you have four single points of failure that happen to be people. Any one of them changing jobs, moving, or retiring takes a quarter of your income with them, and this is the most common way an apparently healthy small business falls over.
There is also a transferability problem, which matters if you ever intend to sell. Revenue that exists because customers like you specifically does not transfer to a buyer, which is the same mechanism I described in the business elements that have evidence behind them under owner dependency. Relationship revenue is the most durable kind while you are present and the least durable kind the moment you are not.
None of this argues against relationships. It argues against relying on them without noticing what you are relying on.
Difficult periods are where relationships are actually formed
Here is the part most writing on this subject misses, and I am labeling it as reasoning rather than research, because I have not found a study on it worth citing.
Normal conditions tell you almost nothing about anybody. Everybody is reliable when the work is straightforward and the invoice gets paid on time. A relationship under easy conditions is an assertion, not evidence, which is why years of pleasant quarterly lunches can leave you with no real information about whether somebody is any good to work with.
Something going wrong is the only time information gets exchanged in both directions. A project runs late. A supplier fails. Money gets tight and somebody has to wait. In that window you learn who tells you early, who absorbs the part that is theirs, who stays reachable, and who goes quiet. They learn the same about you, and that is the asymmetry worth understanding, because a bad stretch handled well builds a stronger relationship than a good stretch does.
This is the same logic as third party proof outranking your own testimonials. Conduct under pressure is evidence. Everything else is description.
Four things are what people actually remember, and none of them require the problem to be your fault.
Telling them before they find out. The failure is rarely what damages the relationship. Learning about it late is. A message saying this is going to slip and here is why, sent before anybody asks, is worth more than a recovered schedule.
Naming the new date and holding it. Vague reassurance reads as avoidance. A specific revised commitment, met, rebuilds more than the original one would have.
Absorbing the share that is genuinely yours. Not all of it. The part that is yours, visibly, without being asked, and usually it is cheaper than it feels.
Not disappearing. This is the one that does the real damage. People forgive the problem and do not forgive the silence, because the silence is a statement about how much they matter when being useful to them is inconvenient.
Burning a bridge costs more than it looks like it costs
The usual argument here is about reputation and it is true as far as it goes. There is a structural argument that is stronger.
A bridge is not a person. It is an access point to an information neighborhood, which is the whole point of the research at the top of this piece. Burn it and you do not lose one contact. You lose every path that ran through them, including the paths you do not know exist, and those unknown paths are where the valuable weak ties live.
Industries are also far smaller than they look from the inside. The contractor you cut off and the client you want next year share a connection you cannot see, and in a city like Las Vegas, where most trades operate in overlapping circles, that overlap is not a hypothetical. The damage then spreads along exactly the channels that produce opportunity, which is the worst possible routing.
The reason people keep doing it anyway is a timing problem. The satisfaction of saying the thing is immediate. The cost arrives two years later as a call that never comes, detached from its cause, so you never find out you paid it. A cost you cannot observe is a cost you never learn to avoid.
One honest qualification, because the advice is usually delivered as never end anything. Some relationships should be ended. A client who does not pay, a partner who misrepresents you, anybody who treats your staff badly. Keeping those is not generosity, it is a tax you pay repeatedly.
The distinction is between ending a relationship and burning it. Ending it is declining the next project, letting a contract lapse, saying you are not the right fit for this. Burning it is leaving a story behind that gets retold without you present. You can do the first without the second, and the first is often necessary.
What makes the difference is almost entirely the exit. Last impressions get repeated far more than first ones, because they are the interesting part of the story. The handover done properly on the way out, the final invoice you did not pad, the referral you made to a competitor who was a better fit for the job. Those get mentioned for years, and they get mentioned by somebody with no reason to flatter you, which is the only kind of recommendation that carries weight.
What to actually do
Four practices, in order of how much they return relative to effort.
Maintain the wide layer deliberately, because it decays by default. The close relationships maintain themselves, since you will see those people regardless. The two hundred moderate acquaintances are the ones that quietly go cold, and those are the ones the research says matter most. This is the opposite of where attention naturally goes.
Make the contact cheap and free of an ask. The reason the wide layer decays is that reaching out feels like it requires an occasion. It does not. Sending somebody a thing they would find useful, with nothing attached, is sufficient, takes two minutes, and is the entire practice. A relationship that only ever produces requests stops being one.
Exit everything cleanly, including what you end. This is free and almost nobody does it, which makes it the highest return item on the list. Every relationship you leave well is a bridge that still carries traffic, and every one you leave badly removes an entire neighborhood of connections you will never be able to count.
Count your concentration once a quarter. What share of revenue came through your top three relationships. If it is above half, that is your most urgent structural problem regardless of how good this quarter looked, and the fix is not better relationships. It is a source of customers who found you without one.
The honest summary
Relationships matter enormously, and the way that sentence is usually meant is wrong in two directions.
It overvalues closeness, when the evidence points toward the moderate acquaintances you rarely speak to as the ones that generate opportunity. And it undersells the risk, because a business built entirely on personal relationship has a hard ceiling, a concentration problem, and nothing that survives the owner stepping back.
It also mistakes where relationships get built. Not over lunches in good quarters, but in the weeks when something has gone wrong and people find out what you do under pressure. That is the only time either side learns anything real.
The practical version is unromantic. Keep a wide layer of weak connections warm at low cost, handle your bad weeks in the open, leave cleanly even when you are the one leaving, and make sure some of your customers arrive without knowing anybody at all.