How to Spot a Bad Freelance Client Before You Sign
By RemoteSWE ·
Short answer: almost every client who ghosts, disputes a payment or grinds you down on scope showed at least two warning signs before any work started. The signals are boring and consistent — no budget named, urgency without a deadline reason, a request for free sample work, resistance to a deposit, and vagueness about who actually approves things. None of them is disqualifying alone. Two or more together is a pattern, and the cost of walking away is a few hours of your time versus weeks of unpaid work. This post gives you a ten-point rubric to score a client before you sign, plus the green flags worth chasing.
The short answer: score the client out of 10
One point for each item that is true. Score before you quote, not after.
| # | Warning sign | Why it predicts trouble |
|---|---|---|
| 1 | Will not name a budget range | Either there is no budget, or the plan is to anchor on your number |
| 2 | Urgent, but cannot say what the deadline is tied to | Manufactured urgency is used to skip the paperwork |
| 3 | Asks for free sample work or a "small test task" | Paid trials exist. Unpaid ones are a business model |
| 4 | Pushes back on any deposit | They want you to carry all the risk |
| 5 | Cannot say who signs off | Approval will move every time you reach it |
| 6 | Brief is written entirely in adjectives | Nothing countable means nothing enforceable |
| 7 | Mentions the last freelancer was terrible | Sometimes true. Often the third time this year |
| 8 | Wants to move off-platform immediately | Removes escrow, dispute process and paper trail |
| 9 | Negotiates the rate before discussing the work | Price is the only variable they care about |
| 10 | Slow and vague pre-sale, but expects instant replies | Whatever happens now gets worse after signing |
0 to 1: proceed normally. 2 to 3: proceed with a deposit and a tightly countable brief. 4+: the expected value of this project is negative. Decline politely and spend the time on outreach.
Why the signs are so easy to ignore
Every warning sign has a completely reasonable innocent explanation, and when your pipeline is thin you will find it. No budget named? They might not know the market. Wants a quick test task? Fair enough, they do not know you either. Urgent? Businesses are busy.
That reasoning is not wrong in any individual case. It is wrong in aggregate, because you only apply it when you want the work. The rubric exists to move the judgement to a moment when you are not yet invested, which is the entire trick — the same client looks very different before you have written a proposal and after you have spent three hours on one.
It is also worth being honest about frequency as a diagnostic. If difficult clients feel like an unlucky streak, count them. One bad account a year is variance. Five is a sourcing problem, and no amount of contract tightening fixes where the clients are coming from — that is a channel question, covered in where to find clients outside the marketplaces.
The four situations that cost the most
The retroactive refund. A completed, paid, approved project, and weeks later a demand for part of the money back — sometimes with a third party brought in to make vague threats about the quality of your work. What protects you here is not a better argument but written sign-offs at every stage, dated.
The delayed chargeback. Paid, delivered, and then months later the charge is reversed at the card issuer. These are winnable far more often than people assume, and they are won on documentation: your agreed terms, evidence of work in progress, and a client email saying the work was approved.
The slow squeeze. Nothing dramatic. A small extra request every week until the project contains twice the work it was priced for. This one is prevented by process rather than by client selection — see how to stop scope creep.
The disappearance. Final files sent, then silence. Almost entirely preventable by changing the order in which you deliver and invoice, which is covered in what to do when a client ghosts after delivery.
A worked example: two enquiries, same week
Enquiry A: "Hi, we need a developer urgently for an exciting project. Our last developer let us down badly. Budget is flexible for the right person — can you send your best price? Also, could you do a quick sample screen so we can see your style? We move fast and need someone responsive."
Scores 5: no budget, manufactured urgency, free sample requested, previous freelancer blamed, price-first framing. Every sentence is warm and every sentence is a flag.
Enquiry B: "We run a 12-person clinic and our patient follow-up is manual. Budget is around $6,000 to $9,000 and we would want it live before the new year because our current system contract ends in January. I would be approving it, our practice manager would be testing. Do you take a deposit, and what do you need from us to scope it properly?"
Scores 0, and note what it contains: a range, a deadline tied to a real event, a named approver, and a client who raised the deposit before you had to. Enquiry B is worth reorganising your week for. Enquiry A is worth a polite decline, even in a quiet month.
Green flags worth chasing
- They name a budget range unprompted. Usually means they have bought this kind of work before.
- The deadline is tied to something real. A contract expiry, a launch, a season — not "as soon as possible".
- They ask how you like to work. Process questions predict a process-respecting client.
- They raise payment terms first. Businesses that pay people are comfortable talking about paying people.
- A named decision-maker who is in the conversation. Approval will not wander.
- A written brief with countable things in it. Screens, users, endpoints, locations.
- A hire history you can check. On a marketplace this is visible — hire rate and spend are the strongest single filter, as covered in why your proposals are not being viewed.
How to decline without burning the lead
Most bad-fit clients are not bad people, and some of them become good clients later under different conditions. Decline in a way that leaves that door open.
- Decline on capacity or fit, not on their behaviour. "I do not think I am the right fit for this one" needs no elaboration and invites no argument.
- Offer the version you would say yes to. "I could do a scoped first phase at $X with a deposit" converts a decline into a counter-offer. A surprising number accept it.
- Do not free-consult your way out. Long explanations of what is wrong with their brief are unpaid work.
- Log it. One line, with the score. Patterns in who you decline tell you as much about your channel as patterns in who you sign.
Declining is a skill with compounding returns: every hour not spent on a four-flag client is an hour spent on outreach that produces a zero-flag one.
FAQ
How do I know if a client is legitimate?
Check what you can verify rather than what they tell you: a real company presence, a named individual who is actually in the conversation, payment verified if you are on a marketplace, and a hire history you can look at. Then apply the rubric. Legitimacy and workability are different questions, and the second one causes more losses.
Should I ever do a free test task?
No — but paid trials are reasonable and often a good idea for both sides. If they will not pay for a small scoped piece of work, the objection is not about assessing your skill.
Is it normal for clients to be this difficult?
Some friction is normal; a steady stream of disputes, refunds and disappearances is not. If it feels constant, count the cases over twelve months. The number usually points at where the clients are coming from rather than at freelancing in general.
What if I need the money and the client scores 4?
Then reduce your exposure instead of ignoring the score: take the whole fee up front, cut the project into a small paid phase, and define the deliverables as countable objects. If none of those are acceptable to them, the score was telling the truth.
Score the client before you write the proposal, while you are still capable of being objective about it. No budget, false urgency, free samples, deposit resistance, no named approver — two of those is a pattern and four is a decision. Then protect the ones you do take with the boring mechanics: countable brief, deposit, written approvals at every stage, and files released on payment. Those same records are what win a chargeback if it ever comes to that. Start with the payment ladder and the change-request habit, and if bad clients feel constant, fix the channel using where to find clients off-platform. If they are coming from Upwork, start by checking what your profile is signalling with the free profile scorer.