How to Find Work Without a Network: What a 20-Million-Person Experiment Revealed
Every piece of advice about how to find work without a network ends in the same place: build one. Go to events. Make connections. Nurture relationships. Come back in a year.
Which is useless when you need work now, and it turns out to be the wrong instruction anyway.
In 2022, researchers published the result of something nobody had managed before — a genuine controlled experiment, on 20 million people over five years, testing whether the connections people assume matter are the ones that actually produce work. The finding contradicts the advice, and the half of it that almost never gets quoted contradicts the popular version of the finding too.
Meanwhile, government data shows the channel with the strongest evidence behind it is the one fewest people use.
The Channel Almost Nobody Uses
Start with what people actually do, because it sets up everything else.
The U.S. Bureau of Labor Statistics tracks which methods unemployed jobseekers use. For 2025, the breakdown looks like this:
| Method | Share of jobseekers using it |
|---|---|
| Sent out resumes or filled out applications | 56.1% |
| Contacted employer directly | 50.8% |
| Contacted friends or relatives | 23.7% |
| Public employment agency | 13.8% |
| Private employment agency | 8.4% |
The average jobseeker uses 1.8 methods, which is why the column adds to more than 100%.
Look at the third row. Fewer than one in four people reach out to anyone they know. The overwhelming majority are sending applications into a system where they compete anonymously against everyone else doing the same thing.
And the waiting is not short. In September 2026, the median spell of unemployment ran 11.5 weeks, the average 24.8 weeks, and 27.1% of unemployed people had been looking for 27 weeks or more. The gap between the median and the average is the tail — a substantial group waiting far longer than the typical figure suggests.
What the Experiment Actually Did
The idea that loose acquaintances matter more than close friends is not new. Mark Granovetter proposed it in a 1973 paper in the American Journal of Sociology, arguing that close contacts tend to know the same things you do, while distant ones reach information you would never otherwise encounter.
For fifty years that remained a correlation. People who got jobs through acquaintances might simply have been different people.
Then a team from LinkedIn, Harvard, MIT and Stanford — Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson and Aral — ran the experiment, published in Science in September 2022.
They manipulated LinkedIn’s “People You May Know” algorithm. Different variants pushed users toward forming more weak ties or fewer of them, at random. Over five years this covered more than 20 million people, roughly 2 billion new connections, and about 600,000 new jobs.
Randomising who gets which kind of connection removes the problem that ruined every previous study. The differences in outcome are caused by the tie, not by the person.
The Half of the Finding Everyone Skips
The headline result is that weak ties increase job mobility. That part got widely reported.
Here is the part that did not, and it changes what you should actually do: the relationship is not linear. The researchers found that weaker ties increased job transmission only up to a point, after which there were diminishing returns to weakness. The shape is an inverted U.
Sit with what that means. Ties that are too close do not help much, because they already know what you know. But ties that are too distant stop helping as well. There is a middle zone, and that is where work actually comes from.
| Type of connection | What it gives you | How useful |
|---|---|---|
| Close friends and family | Support, and information you already have | Low for finding work |
| People you know slightly — former colleagues, people met once, dormant contacts | Access to a different information pool, plus enough context to vouch for you | Highest |
| Complete strangers | A different pool, but no basis to recommend you | Falls off again |
That middle row is the practical answer, and it is almost never what networking advice describes. It is not about meeting new people and it is not about deepening friendships. It is about a category of person you already have and have been ignoring.
Why This Matters More If Your Work Is Digital
The study found something else worth knowing: the effect varies by industry.
Weak ties benefited workers in more digital industries, while stronger connections mattered more in less digital ones. Information about digital work travels well through loose, distant links. Information about physical, local work apparently does not.
If you are working in anything that produces output over the internet — writing, design, development, analysis, marketing, support — you sit squarely in the segment where the weak-tie effect is strongest.
That segment also happens to be where most people work alone. For a business of one, with no colleagues and no employer passing opportunities down, the dormant-contact channel is not one route among several. It is most of the pipeline.
That segment is large but far from universal. BLS data shows 37% of wage and salary workers could work at home and 34% did so at least occasionally. So this applies to roughly a third of the workforce, concentrated in exactly the kind of work that is most exposed to being reshaped by automation and most dependent on trust and relationships to survive it.
How to Find Work Without a Network You Think You Have
The useful reframe is that you are not missing a network. You have one, made of dormant connections you have written off because the relationship is not close.
Those are precisely the connections the data says work best.
List the dormant ones, not the close ones. Former colleagues. People from a course or a job two roles ago. Someone you worked with on one project. Someone you met once at an event and exchanged messages with. If you can name how you met and they would recognise your name, that is the zone.
Reactivate rather than cold-approach. The distinction matters because the study’s inverted U suggests the value sits where some shared context survives. A message to someone who remembers you is a different object from a message to a stranger, even when the two messages read identically.
Ask for information, not a job. “Do you know how hiring is going in your area?” and “Who should I be talking to about this kind of work?” are answerable. “Do you have a job for me?” is not, and it puts the other person in the position of saying no.
Do it at volume, because it is probabilistic. No individual message is likely to produce work. The mechanism is that information reaches you from a pool you do not otherwise touch, and more connections means more pool. Twenty messages is a reasonable experiment. Two is not.
Keep sending applications anyway. The research says the network channel has causal effect. It does not say the other channels have none.
Why “Build Your Network” Is the Wrong Instruction
Three reasons the standard advice fails in practice, now that we know the shape of the curve.
It targets the wrong people. Networking events produce strangers, which the inverted U says is the far end where returns fall off again. Reactivating a dormant contact costs nothing and lands closer to the optimum.
It is slow in the wrong direction. Building relationships takes months and moves ties toward the strong end — away from where the evidence says the value is. You can spend a year converting useful weak ties into less useful strong ones.
It assumes the problem is quantity. For most people the problem is that existing connections are unused. Fewer than a quarter of jobseekers contact anyone they know at all. The constraint is not supply.
None of this means relationships do not matter. It means the relationships that produce work and the relationships that produce closeness are not the same set, and treating them as one is why the advice disappoints.
What You Bring to the Conversation
A weak tie can pass you information. It cannot make you employable.
The reactivation only converts into work if there is something on the other end — a skill someone pays for, delivered at a price you can state. What the labour data actually rewards is less exotic than most career advice suggests, and worth being honest with yourself about before you start sending messages.
The same applies to rate. If the conversation goes well, someone will ask what you charge, and the answer should come from published wage data rather than from what feels defensible in the moment. Being discovered through a weak tie and then underpricing yourself by half is a specific and very common way to waste the opportunity.
Where This Does Not Apply
Three honest limits, because the study is narrower than the way it tends to get summarised.
It measured jobs, not freelance clients. The outcome variable was job transmission on LinkedIn. The underlying mechanism — information about opportunity travelling through loose links — plausibly extends to client work, but the experiment did not measure that, and I am not going to claim it did.
It measured LinkedIn. One platform, with its own population and its own norms. The effect in other settings is unmeasured.
It is about information flow, not about being chosen. A weak tie improves the chance you hear about something. Everything after that is still the ordinary process of being evaluated, and nothing in this research changes that part.
Frequently Asked Questions
What counts as a weak tie? Someone you have a real but inactive connection to — a former colleague, a classmate, someone from a previous role or a single project, an acquaintance you could name the context for. Not a close friend, and not a stranger. The research found the strongest effect in the middle of that range rather than at the extremes.
Isn’t this just saying to use LinkedIn more? No. The study was run on LinkedIn, which is where the data came from, but the mechanism is about where information travels, not about one platform. The same dormant contacts exist in your email, your phone, and your memory.
How many people should I contact? Enough that the probability works in your favour. Each individual message is unlikely to produce anything; the effect comes from reaching a pool of information you otherwise would not touch. Twenty is a reasonable minimum to learn anything from the attempt.
What do I actually say? Ask for information rather than for a job. Questions about how their area is hiring, who else you should be speaking to, or what is changing in their part of the industry are answerable. A direct request for employment forces a yes or no, and the default answer is no.
Does this work if my industry isn’t digital? Less well, according to the study. The researchers found weak ties benefited more digital industries while stronger ties mattered more in less digital ones. If your work is local and physical, closer connections are likely the better route.
How long does finding work usually take? BLS data for September 2026 puts the median spell of unemployment at 11.5 weeks and the average at 24.8, with 27.1% of unemployed people searching 27 weeks or longer. The gap between median and average shows how long the tail is.
Should I stop sending applications? No. More than half of jobseekers use applications and direct contact with employers, and the research does not say those channels fail. It says the network channel has a measurable causal effect and is badly underused.
Is the research solid? It is a randomised experiment published in Science, covering over 20 million people across five years, which is an unusually strong design for a question of this kind. The main caution is scope: it measured job mobility on one platform, not every form of finding work.
Conclusion
The question of how to find work without a network rests on a premise that is usually false. Almost nobody has no network. What most people have is a set of connections they have discounted because the relationship is not close — which is exactly the category the evidence points to.
The experiment on 20 million people found that weak ties causally increase job mobility, with two qualifications that matter more than the headline. The relationship is an inverted U, so connections that are too distant stop helping just as connections that are too close never helped much. And the effect is strongest in digital industries, which is where most remote and independent work now sits.
Set against that, government data shows fewer than a quarter of jobseekers contact anyone they know, while over half send applications into a queue.
The gap between those two facts is the opportunity, and closing it does not require building anything. It requires a list of people you already met, and a message that asks for information rather than for a favour.
We’re also preparing a more detailed course walking through the exact systems used to build and monetize this site, step by step. Readers of these articles will get early access when it’s ready.
A Note from Lucas
I went looking for this research because “build your network” is the advice I have found least actionable in my own working life, and I wanted to know whether it was advice or folklore. It turns out to be a real effect measured properly — and the part that genuinely changed my mind is the inverted U, which I had never seen mentioned in any of the dozens of articles citing this study. Everyone stops at “weak ties are better” and skips the finding that ties can be too weak to help. My honest reading is that most of us are sitting on a list of former colleagues we have decided it would be awkward to message, and that list is the asset. I wrote the section about asking for information rather than for a job because that is the mistake I made the last time I tried this, and the silence afterwards taught me more than the research did.
About the Author
Lucas Palma is the founder and editor of Info Is The Key. He writes about online business, entrepreneurship, and the practical mechanics of building income independently, working from primary sources and official data rather than recycled advice.
Last Updated: October 2026 | Reading Time: 10 minutes
Editorial note: This article summarises published peer-reviewed research and U.S. government statistics as of October 2026. The experiment described measured job mobility on one professional platform and does not establish results for freelance client acquisition, other platforms, or other countries. Labour market figures are United States data and will differ elsewhere. Nothing here guarantees an employment outcome.
Sources
- Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson & Aral — A Causal Test of the Strength of Weak Ties, Science (September 2022)
- Granovetter, M. — The Strength of Weak Ties, American Journal of Sociology (1973)
- U.S. Bureau of Labor Statistics — Unemployed jobseekers by active job search methods used
- U.S. Bureau of Labor Statistics — Unemployed persons by duration of unemployment
- U.S. Bureau of Labor Statistics — Job Flexibilities and Work Schedules
2 Comments