Should you specialise? The generalist question, answered with evidence
Specialise — but later, and on evidence. A clear niche usually raises pricing power and makes referrals easier to give, because clients can describe you in one phrase. It also shrinks the pool of work you are visible for, and the premium arrives months after the generalist enquiries slow down. So the sequence matters more than the choice: earn as a generalist first, watch which work proves profitable, repeatable and referred, then narrow towards that overlap while the general work still pays the bills. This guide covers what a niche actually is, the case for and against, the evidence test, and how to narrow without burning the income that funds the change.
What a niche actually is
A niche is not a slogan, a new logo or a line in a bio. It is a decision about which work you are visibly for — and it can be made on three independent axes. The trade axis: one skill, offered to anyone — the bookkeeper, the motion designer. The industry axis: your usual skill, aimed at one kind of client — a web developer who builds only for dental practices, a writer who works only for fintech firms. The problem axis: one outcome, however it gets delivered — "I fill empty appointment books", "I rescue product launches that stall".
Most freelancers only need to narrow on one axis. Narrowing on all three at once — one skill, one industry, one problem — leaves a market too small to find you. One axis, clearly stated, is what makes a stranger able to say who you are for.
The trade axis is the one you can watch being priced. The rate calculators on this site run the same maths for forty trades, and the spread between them is not explained by effort. The virtual assistant rate calculator works out at $58 an hour; the brand strategist rate calculator works out at $152 an hour. Same formula, same leave and tax assumptions. The difference is scarcity and legibility: how many people can credibly offer the work, and how clearly a client can picture what they are buying. Run your own trade through the hourly rate calculator and you will see your floor. A niche is how you raise the ceiling above it — legibility, chosen deliberately instead of left to chance.
The case for specialising
Pricing power. A client comparing five interchangeable generalists chooses on price, because price is the only difference they can see. A client who needs the person for a specific job compares you against a much smaller set — often against doing nothing at all — and the conversation moves from your rate to your availability. Scarce and legible beats capable and vague, every time the two are priced side by side.
Referrals compound. People refer what they can describe in one breath. "She does websites" competes with everyone the referrer has ever met; "she does websites for physiotherapy clinics, and their bookings went up" travels on its own. Every finished niche project makes the next introduction easier to give.
Delivery gets faster. In a niche, the same problems recur: the same integrations, the same objections, the same three failure points. Faster delivery at the same price is a pay rise that never had to be negotiated — and the tenth version of a problem is usually better than the first.
Marketing gets cheaper to aim. One audience, one vocabulary, one set of examples that all point the same way. A generalist's portfolio has to be re-explained project by project; a specialist's reads as a single proof, repeated.
The case against: what a niche costs
The pond is smaller. That is the point, and also the risk. Visibility narrows before reputation deepens: enquiries from outside the niche fall away months before niche work fully replaces them, and there is a valley in the middle where both are thin.
Boredom is real. The same problems recurring is efficiency from the client's side of the desk and monotony from yours. Some freelancers thrive on repetition refined; others do their best work when every project is strange to them. A niche you resent gets marketed half-heartedly and delivered flatly, and clients can tell. Fit is a business input, not a luxury.
A niche can die. Specialise in one platform, one tool or one algorithm, and your market has a landlord — who can change the rent, the rules or the locks. Trades and problems outlive tools: bookkeeping for cafés survives any accounting package; "expert in this year's software" does not survive the software. When you choose an axis, anchor to the one least likely to be discontinued.
An imagined niche is a guess. The riskiest version is picked from preference alone — the industry you find interesting, the work you wish clients bought more of. Without paid evidence underneath it, a niche is a bet placed with your entire pipeline. The next section is how to place the same bet with the evidence stacked first.
The evidence test: your last ten projects
Take your last ten paid projects — paid is the operative word — and score each one on three questions. Was it profitable at the price charged, once the real hours are honestly counted? Was it repeatable — could you deliver it again, faster, using what you now know? Did it produce, or could it plausibly produce, a referral or a repeat booking? Then look for the overlap: the trade, industry or problem that keeps appearing among the projects that answer yes. One enthusiastic project is an anecdote; three pointing the same way is a pattern.
Specialise towards the overlap, not towards the work you wish you had. The wish list is a research project. The overlap is a market that has already paid you. If the raw material for this exercise lives anywhere, it is in the client list you build for the year-end admin checklist — the year's clients, what each one paid and where each one came from is exactly the dataset this decision needs.
And if fewer than three of the ten point the same way, you do not have a niche yet. You have a direction to test: accept the next few projects with the question in mind, and let the pattern finish forming before you build a public position on it.
Narrow without burning the generalist income
You do not resign from general work. You narrow in public, gradually, while the general work keeps paying — the income you already have is what funds the transition.
Run a 70/30 season. For a season — a quarter is a fair unit — aim most of your new effort at the niche: its own page, its own examples, outreach written for that one reader, and keep the remaining effort on the general pipeline. The niche offer deserves proposal language written for its reader, too — the proposal generator makes it cheap to keep a niche version of your standard proposal alongside the general one. The exact proportions matter less than the direction: niche effort rising, general effort maintained.
Price the general work up instead of refusing it. When off-niche work arrives during the transition, do not turn it away — charge it properly. A higher generalist rate, raised with the notice-and-boundary method in how to raise your rates, does three things at once: it keeps the income, it rations your generalist hours in favour of niche work, and every client who accepts anyway is paying a premium that funds the change. Refusing work you still need is purity. Pricing it is strategy.
When not to specialise yet
Your first year. You have no last ten projects to test — or the ten you have are whoever said yes first. That is the evidence-gathering phase working as intended. Take the range, keep the notes, run the test when there is something to test.
No pattern yet. If your ten projects point in ten directions, the market has not told you anything — it has told you that you are employable, which is a different and also useful fact. More repetitions, wider net, test again. Specialising now would mean choosing by taste and calling it data.
Your local market is too small. If your clients are local and your town has four cafés, "bookkeeping for cafés" is a ceiling, not a niche. Widen one axis — all hospitality businesses, or the cafés of the nearest city — or take the niche remote, where the pond is far larger.
None of these is "never". Each one is "not yet — and here is the evidence to gather in the meantime". The freelancers who get this decision wrong are rarely the ones who moved slowly; they are the ones who moved on a slogan.
Questions
Related: positioning is one lever among several
Positioning sets the ceiling; arithmetic sets the floor. Work your floor out with the hourly rate calculator, trade by trade, and the best AI tools for freelancers pillar covers the kit that wins and runs the work. When the specialist rate is set, it still needs announcing properly — how to raise your rates covers the timing and the email — and the dataset behind the whole decision is the client list in the year-end admin checklist. Choose the niche on evidence, narrow it in public, and let the general work pay for its own replacement.