The phrase "your network is your net worth" has done a lot of damage. It turns something ordinary — knowing people, being known — into a growth metric, and it produces the specific kind of behaviour everyone can smell: the connection request from a stranger, the coffee that turns out to be a pitch, the message that arrives after four years of silence and needs a referral by Friday.
The underlying fact is real, though, and worth stating without the slogan. Hiring runs on incomplete information. A résumé is a weak signal, an interview is a noisy one, and the fastest way for a hiring manager to reduce uncertainty is for someone they trust to say something specific about you. That is the whole mechanism. LinkedIn's own survey of professionals found that 70 per cent of people hired in 2016 had a connection at the company that hired them. It is LinkedIn's data about the value of a thing LinkedIn sells, so hold it loosely — but as a description of how jobs reach people, it matches what I see from the hiring side.
What a referral actually does
Referrals are routinely oversold, in both directions. They do not get anyone hired. They almost always get a CV read by an actual human rather than filtered in a batch, and that is not a small thing — it is often the difference between a real evaluation and none.
But not all referrals are the same, and the difference is visible from my seat. A referral that says "I worked with her for two years, she owned the data pipeline when it was failing weekly, ask her about how she debugged it" changes how I read the application. A referral that says "good guy, worth a look" tells me only that somebody submitted a form, possibly with a referral bonus attached. Referrals carry the credibility of the referrer, which means asking a near-stranger for one usually produces the weak kind.
That has a practical consequence for the person asking. Being able to name two or three people who could describe your work in specifics matters much more than the size of your contact list. If nobody could write the first kind of sentence about you, more networking events will not fix it. Doing visible work with people who see it will.
The weak ties thing, stated carefully
Mark Granovetter's 1973 paper on the strength of weak ties is the origin of the idea that acquaintances, not close friends, are the ones who bring you new opportunities. The logic holds up: your close contacts largely know what you know and hear about the same openings you hear about. Someone two steps away is plugged into a different stream of information.
The useful implication is not "collect acquaintances." It is that the people you talk to twice a year — the former colleague now somewhere else, the person from a conference, the acquaintance in an adjacent field — are structurally more likely to be the route to something new than your immediate circle, and they are the ones most easily allowed to lapse. Keeping a loose connection alive costs very little, and it is the part of a network that actually generates surprises.
Being findable beats reaching out
The people who contact me with something I want to talk about are, almost without exception, people whose work I encountered first. A paper, a repository, a thoughtful answer in a forum, a blog post that got something right. The connection request with a generic note is the weakest possible version of the same intent.
For engineers and researchers this is the highest-leverage move available, and it is available to people with no access and no existing network. Write up the thing you debugged. Publish the tool you built for yourself. Answer the question properly in public. It is slow and it compounds, and unlike attending events, it works while you are asleep and it works for people who find conferences exhausting.
Profile hygiene matters at the margin — a searchable, current, honest description of what you actually do, with a photo. LinkedIn reports that profiles with a photo receive substantially more views, by their own account around twenty-one times more, which is a platform marketing figure rather than an independent finding and should be read as "yes, add a photo," not as a multiplier you can bank on. None of this is where the value is. It is table stakes; the work is the signal.
Asking well
I get a fair number of cold messages. The ones I answer share a shape: short, specific, obviously addressed to me rather than to a list, asking for something bounded, and easy to decline without awkwardness. "I am moving from academia into industry ML and I am stuck on whether to target research or applied roles — would fifteen minutes be possible, or if not, is there something you would point me at?" takes a minute to answer and I usually do.
The ones I do not answer ask for something unbounded: be my mentor, review my career, refer me somewhere you have no basis to vouch for me. Not because the people are unreasonable, but because the request has no edges, and a request with no edges is a request to design the help as well as give it.
Two things make an enormous difference afterwards, and almost nobody does them. Tell the person what you did with what they said — not thanks, but "I took the applied route and here is how it went." It is the only way they learn whether their advice was any good, and it is the single most reliable way to turn one exchange into an ongoing relationship. And when you are asked for something in return later, say yes if you can.
Relationships survive on the years when you need nothing
The failure mode is not rudeness. It is that most professional relationships are maintained only while they are useful, and so they are all in a state of mild disrepair exactly when you need one.
The fix is not a spreadsheet with a contact cadence. Treating people as rows to be touched quarterly produces messages that read exactly like what they are. What works is lower effort and more honest: pass on things that are genuinely relevant to a specific person, congratulate people when something goes well for them, answer questions when asked, make introductions where you can see a real fit. Then let it be quiet for a year without treating that as a problem. Relationships between adults with jobs are allowed to be dormant. They are not allowed to be extractive.
Generosity without scorekeeping works better than reciprocity accounting, and I do not think that is a moral claim so much as an observation about how the score is kept anyway. People remember who was useful to them, and the recall is long.
Mentors: ask for the conversation, not the title
The advice to "find a mentor" is unhelpfully abstract, and the request "will you be my mentor" is hard to accept because it is a commitment to an unspecified amount of future work.
What works is inverting it. Ask one person one concrete question about a decision you are actually facing. If the exchange is useful to both of you, it repeats. If it repeats often enough, it was a mentoring relationship all along and nobody had to name it. Most of the useful mentoring I have both received and given arrived that way rather than through a programme.
One structural point worth knowing: a mentor inside your reporting line and a mentor outside it are different instruments. The ones with influence over your progression are useful for navigating the organisation and useless for the questions that are risky to say out loud. Most people need at least one of each.
What networks do not do, and who this advice assumes
Three honest caveats.
Most networking advice assumes a specific culture — American or Western European, white-collar, comfortable with casual contact across hierarchy, built around conferences and coffee. Plenty of workplaces do not run that way. In many Indian and East Asian organisations, approaching someone several levels senior without an introduction reads as presumptuous rather than enterprising, and the functional equivalents are institutional: alumni networks, professional bodies, former colleagues, an introduction from someone with standing. The mechanism is identical — someone credible vouching for you — but the route is different, and importing the American version wholesale tends to produce awkwardness rather than access.
Second, referral-driven hiring reproduces whatever inequality already exists. If opportunity travels through networks, then people who arrive with networks — the right college, the right city, the right family — get a head start that has nothing to do with capability. Both things are true at once: you should use referrals because they work, and you should notice that a system running on them is not fair. If you are on the hiring side, that is an argument for taking cold applications seriously and for referring people who are not already like you.
Third, a network does not fix a weak case. It gets you read, considered, and occasionally warned about a bad job before you take it. It does not survive you being unreliable, because the only thing a referrer is actually spending is their own credibility, and people are appropriately careful with that. Which is, in the end, the whole of it: be someone specific that someone specific would vouch for, stay reachable, and be useful when it costs you little. Everything else in this genre is decoration.
Tags
Taresh Sharan
support@sharaninitiatives.com