When to say no to a project
The short version: say no when a project fails one of four tests — fit, price, timing or trust — and say it before the contract, while declining is still free. A wrong yes costs far more than the fee it brings in. It takes the billable weeks your rate was calculated on and fills them with the wrong client, and it concentrates your income in work that teaches you nothing, shows in no portfolio and leaves you too busy to take the right project when it arrives a month later. The expensive version of this decision is firing a client halfway through a relationship. Nor is the worst answer a no — it is the slow maybe, which spends the client's Plan-B time, and your own peace of mind, on a decision you have in truth already made; a clean no sent promptly is a professional courtesy. This guide is about the cheap version: the four tests, the red flags, and how to deliver the no so that the door stays open.
1. The fit test
Start with the work itself, before money or timing enter it. The question is whether this project belongs in the business you are building — the direction our guide to choosing a freelance niche sets out. Work well outside that direction is not neutral filler: it occupies the same weeks the compounding work needed, and it produces nothing the next right client can be shown.
The portfolio question. Imagine the project finished, well and on time. Would you put it on your case-study page? Our guide to writing a freelance case study describes what a showable project looks like: a recognisable problem, work you actually did, a result the client will confirm. If the honest answer is that you would hide this one — wrong industry, wrong standard, wrong kind of client — the fit test has its answer, whatever the fee.
Skills you would have to fake are a no. Not a hesitant no — a plain one. Faked competence surfaces mid-project, at the worst possible moment, in front of the client least able to absorb it. Skills you would have to learn are a different case: unfamiliar work can be priced honestly, at pessimistic hours, with the learning curve named in the proposal, or carved out as a small paid discovery phase first. Our guide to pricing a project you've never done works through both. The line is not “have I done this before?” — it is “would I have to pretend?”
2. The price test
The fee is not the offer. The offer is the fee minus your hours at your rate, minus anyone you would have to pay, minus the project's own expenses — and that arithmetic takes two minutes before you answer, not two weeks into the work. Run the quote through the project profitability calculator while the enquiry is still in your inbox: fee, realistic hours with a buffer, subcontractor costs, direct expenses. The verdict it returns — profit, or “you pay to work” — is the real content of the client's offer.
A fee that only works if nothing goes wrong is a no at that fee. Projects always go slightly wrong: a revision round nobody scheduled, a client who answers in days instead of hours, a file that arrives in the wrong format. A quote with no slack in it has priced a project that does not exist. What remains, when the number comes back thin, is not automatically a no — it is a no to that scope at that fee.
Which is where the counter-offer lives. Do not move the rate; move the shape: the same fee for a smaller deliverable, the full deliverable at the honest fee, or the full scope in two phases with the second priced later — the same shapes your payment terms should already describe, which is why the terms are written down before a client tests them. Our guide to freelance rate negotiation works through the trades in detail. If the client refuses every reshaping — full scope, full speed, cut price — the no writes itself.
3. The timing test
Your calendar is part of the price, though it never appears on the quote. A yes that needs the weeks you promised to rest, or the weeks already sold to another client, is not really a yes — it is a swap, and swaps should be made with both sides visible. Name what the project displaces before you agree to displace it.
Rest counts as a commitment here, not as empty space. The vacation weeks in your year are already priced into your rate — the rate calculator's arithmetic assumes them — so giving those weeks to a project quietly breaks the maths for the whole year, not just for that month. Our guide to taking time off as a freelancer treats the break as a booking like any other. A project that can only happen inside it is asking you to work your holiday at no premium.
Urgency itself is not a failed test — it is a different product. Rush work, priced at rush rates with the fee doing the apologising, is a legitimate answer, and our guide to rush fees and urgent-work pricing sets out how. What fails is the combination: urgent, large and cheap, all three at once. Two of the three can usually be negotiated. Three is a description of how the project will treat you throughout.
And when the calendar alone fails — the work fits, the client fits, only the weeks do not — the answer is not this guide's flat no but a dated one: saying you're booked out and running a waitlist turns the refusal into a queue, with an honest date the client can plan around.
4. The trust test
The last test is about the client rather than the project, and it runs on patterns any experienced freelancer will recognise — offered here as common practice, not statistics. One flag can be inexperience. Two or three together are a description of the engagement you are being invited into.
The deposit that never lands. Enthusiasm in the messages, silence at the invoice. A standard deposit, asked for plainly, is the cheapest trust test there is: our guide to freelance deposits and upfront payments covers the amounts and the wording. A client who will not commit a fraction before work starts is telling you how the full amount will go at the end.
No written agreement. “We don't really do contracts” means every later disagreement is settled by memory and leverage. The agreement need not be long — our freelance contract basics keeps it to the clauses that matter — but a refusal of any writing at all is a flag, not a style.
“Exposure” as payment. Visibility, future work and a share of an unlaunched idea are not a budget; they are the absence of one, dressed up. Exposure cannot be spent, and it is rarely offered by the clients whose audiences actually buy.
Contempt for the last freelancer. Listen to how they describe the person you are replacing. Some predecessors deserve it; a client who narrates every past hire as incompetent is showing you the review you will one day receive — probably mid-project, to your replacement.
The practical comfort: the two ordinary asks — a deposit, a short written agreement — filter most of these before they cost you anything. A client who refuses both has answered the trust test for you.
For a returning client, the test is a record, not an impression. Your client list audit holds it: the payment history, the scope habits, the revision rounds. A client who pays late every time has already sat this test, repeatedly, and failed it in writing. The no you send them is the one you owe your past self — the version of you who absorbed the last three late payments and quietly promised it would be the final one.
5. How to say it
The graceful no has four parts, and none of them is an apology spiral. Thank them for thinking of you. Give one honest reason in one line. Leave a door open — a later date, a smaller offer, or a colleague's name. And send it promptly — inside two working days of deciding (the working days calculator counts them the way a client would): a fast no is a kindness, because it returns the client's time to find someone else while their deadline is still achievable. The slow no, dragged out by discomfort, is the only genuinely rude version. Email is the right medium for it, too — brief, dated, and kind on paper; for a client you already work with, the same brevity by phone is a courtesy, never a requirement. At 52 words, the whole message looks like this:
The no with an alternative. Sometimes the project fails but a smaller thing passes: the audit instead of the rebuild, the first phase instead of the whole programme. Declining the project while offering the piece you can do — happily, at a stated price — turns a closed door into a narrower open one. Write that smaller scope down properly in the scope of work generator, or price it quickly with the estimate generator, so the alternative is a real offer, not a vague maybe. And if you pass on a colleague's name instead, tell the colleague first: a recommendation should never arrive as a surprise to the person recommended. If you cannot name a person, name the kind of provider instead — the specialism, or the marketplace where such people are found — and say plainly that the fit matters more than any name. Never invent a person to sound helpful: a fabricated recommendation is the one version of this email that can actually damage you.
The fit no. When the work itself is wrong, the reason is the kindest true one available — and it names where the client should look instead. At 75 words:
The terms no. When the project is right and only the terms fail, the email carries the priced alternative — both offers in one breath, no apology attached. At 89 words:
And if you need the money? Then this guide admits its hardest case plainly: sometimes the right answer is a short wrong project, at a fee that survives the price test, taken knowingly and given an end date. What even a tight month cannot afford is the long wrong project — the one that fills a quarter, blocks the right work and quietly renews itself. The structural fix is the runway: a month or two of essentials in reserve turns most forced yeses back into choices, which is what the emergency fund calculator is for. The no gets easier every month the fund grows.
The guilt, examined for a moment. Most reluctance to send these emails is not kindness — it is the fear of being thought difficult. Set it against what the no actually protects: the weeks already promised to the clients on your books, whose work is juggled the way our guide to managing multiple clients at once describes, and who never consented to share you with a project you took out of guilt. And the arithmetic is colder than the feeling: every yes to the wrong work is a no to the right work that arrives next month — run the year through the pipeline income forecaster and you will find it has no line for regret. The clients worth keeping respect a fast, clean no; several of them will send the next enquiry themselves.
6. When the no gets pushed back
More money arrives. This is the discount conversation arriving from the other direction — our guide to handling a client asking for a discount works through the mirror image. If terms were the only failure, your email already named the priced yes, and a client who meets it has simply accepted your offer: take it in good faith. If the failure was fit or trust, a larger fee repairs neither — it only raises the price of the same bad project.
The guilt-trip. “After all the time we spent talking…” is a request to re-open a decision, not new information. Restate the no once, in one line, with the door exactly as open as it was — then stop. A third and fourth reply are not politeness; they are the negotiation continuing, and every round teaches that your no is an opening bid.
Someone sent them to you. If the enquiry arrived through a referrer, the referrer gets one line of thanks — the referral was appreciated, the project was not a fit, you would be glad to be thought of for [your kind of work] — and nothing else. Do not relitigate the decision with the person who vouched for you, and do not let their embarrassment become your yes.
Questions
Related: the expensive version, and where the tests get their facts
If the project is already running and going wrong, the same decision has a second half: our guide to firing a client professionally covers the notice, the handover and the final invoice. The tests get their information upstream — the discovery call script is where fit, budget and red flags surface while asking is still free. The flags themselves get a page of their own: how to spot a problem client before you sign pairs each one with its innocent explanation and the probe that separates them — the read that feeds the trust test above. The limiting case of the price test — a fee of zero, dressed as a spec piece, an exposure offer or an unpaid trial — has its own guide: how to handle free work requests. Saying no well is a small skill. It is also the one that decides what your working year is made of.