How to Fire a Client Too Big to Lose: 5 Practical Rules

Nobody looks up how to fire a client until they already know the answer. I’ve worked with over 400 clients and been fired a lot. No relationship lasts forever, let’s be honest, and it doesn’t get easier. I get depressed every time, even when I know it’s coming.

So I’ve sat in both chairs, and most advice on how to fire a client irritates me for the same reason. Eight signs. Twelve red flags. Then a script. All of it circling what nobody says plainly: the decision is simple, and the timing is hard.

Here’s my answer. There is one line, it has nothing to do with money, and once it’s crossed you don’t try to fix it. You end it as quickly as you can end it well.

The useful question isn’t how many warning signs you can count. It’s what the client is displacing. For me that’s a good employee. Work alone and it’s your own capacity, which is a different calculation and not one I’m qualified to do for you. Brianna Lane is, so she has the middle of this article.

How to fire a client: where the line actually is

For me, how to fire a client comes down to one line: how they speak to you and how they treat your team.

It’s never the hours, the emails at 11pm, or the fee. A difficult client who is demanding and has high expectations has every right to be that way. A client who is rude or caustic cannot be tolerated. Every list on how to fire a client lumps those together and they are not the same problem.

I’ve worked so hard to build a small but mighty team of exceptional people that I can’t risk losing them over a bad client.

If they’re treating my employees in a way that affects their desire to work at my company, that client has got to go. Revenue can be replaced a lot easier than exceptional teammates can.

Work alone and it’s the same question with you as the person being protected. The SBA’s Office of Advocacy counts 36.2 million US small businesses, 82.3% of them with nobody on payroll. For a lot of owners there’s no teammate to protect. You’re the one the rule has to protect, and nobody else will. Plenty do: in a 2023 survey of 1,000 web designers commissioned by Squarespace, in a sample that skewed solo, 57% said they’d fired a client or refused further work.

What about the ones who don’t pay?

Easy one, and how to fire a client is the wrong frame for it entirely.

Clients that don’t pay or are habitually late, you can have the conversation, but usually nothing ever changes. I don’t want to chase their money or pester them. If they don’t value our work and don’t pay on time, we don’t renew.

Note what that isn’t. Not a confrontation, and not how to fire a client in the dramatic sense. It’s a renewal you quietly decline. Every scope of work we write has termination terms, because nothing lasts forever, and the quiet non-renewal is available more often than people realize.

Clients that pay early are the best, and we have some that pay the day after invoicing. If late payment is the symptom rather than the disease, unpaid invoices is the better read first.

Should you raise the price instead of firing a client?

Here I part company with almost everyone writing about when to fire a client, so let me be exact about it.

Ask anyone how to quiet-fire a client and someone will tell you to price them out instead: raise the number until either they leave or the money makes it worth it. I don’t do that. If the problem is behavior, a higher rate doesn’t fix the behavior, it just sets a price on being treated badly.

The same goes in the other direction, and I’ve had it tried on me. Once or twice a client has groveled on the way out and asked, “Well, can we increase our fees?” You hold firm and say, “No, that’s not what this is about.” If money could have fixed it, it wouldn’t have got this far.

If the problem is how somebody treats your people, a bigger number doesn’t fix it. It puts a price on something that shouldn’t have one, and you’ve agreed to be treated that way for money.

A scope problem and a behavior problem look identical from the outside, and how to fire a client depends entirely on which one you have.

Scope problems are fixable and repricing is the right tool. If a client is overly demanding and sucking up our hours, we force a repapering conversation and update the scope of work, with the fee moving to match. Not punishment, accuracy. Same with revisions: two rounds included, charged after that, and the fear of the fee usually prevents the third. If that’s your situation, how to calculate a price increase and handling out-of-scope requests will do more than this will.

Behavior problems aren’t fixable, and that’s when how to fire a client stops being hypothetical. People don’t change, and clients are people, therefore clients don’t change either. If they’re terrible on day one, they’re always going to be terrible. You won’t change them, so accept that, have the difficult conversation now, and get out as quickly as you can.

That’s the rule. I don’t always follow it. Looking back, I usually knew it in my guts from day one and signed anyway, because I was desperate for another account, or busy and didn’t want to give it my full attention. Honestly, I avoid conflict. Twenty-plus years in, I’m still learning it’s easier to have the tough conversation now than a tougher one later. Knowing your line and acting the day somebody crosses it are two different skills.

Can you afford to fire a client right now?

When I ran my bookkeeping practice, this was the assumption that broke most often. “Get out as quickly as you can” takes for granted that there’s something behind the client. If one account is forty percent of your revenue, there isn’t, and no amount of resolve changes the arithmetic. So before you decide how to fire a client, get the number.

Pull the last twelve months. Work out what each client paid as a share of the total, from the bank statement rather than the invoices. Look at the biggest one.

You’ll have read that no client should be more than 10% of your revenue, usually attributed to nobody in particular. Ten percent isn’t a magic limit, but it isn’t arbitrary either. Under the Financial Accounting Standards Board‘s segment reporting rules, a public company discloses the existence and revenue contribution of any customer accounting for 10% or more of total revenue, though not necessarily that customer’s name. The stated purpose is to signal concentration risk to investors.

Treat it as a warning level, not a rule. It’s the level at which the profession decided concentration risk becomes material enough to declare. Your own safe number depends on your margins, your reserves, and how fast you could replace the work. What’s worth noticing is that at forty percent, nobody is declaring anything on your behalf.

Revenue share is only the first number

Thirty percent of revenue is not thirty percent of profit. A demanding client eats labor a well-behaved one doesn’t: rework, waste, rush costs, hours nobody bills for. I saw accounts that were a quarter of revenue and almost none of the profit, protected fiercely because revenue was the only figure anyone had looked at.

When I ran my practice, owners could tell me exactly what their largest client paid and not what that client consumed. The invoice was visible. The extra calls, the revisions, the delayed work and the Sunday-night dread were not. The textbook answer is to track hours for a month and then decide, which is an instruction to postpone by a month. It’s usually been postponed a year already.

So estimate instead. Forty minutes gets you close enough to act on, or to know you can’t yet.

1. What each client paid over three months, from the bank statement.
2. Roughly how many hours each one took. Estimate against yourself: if you think ten, write twelve. You’re forgetting the long calls and the second round of revisions.
3. Payment divided by hours. That’s your effective rate for that client, which is not the same as profit but is close enough to compare one client against another.
4. Your best client’s rate beside your worst client’s rate.

Count the hours that never reach an invoice, because those get undercounted every time.

If the rates look thin across your whole book rather than on one client, you have a pricing problem rather than a client problem, and small business profit margins is the better starting point.

If the number says not yet

Then how to fire a client isn’t the question in front of you. Getting the share down until that conversation is survivable is.

How to fire a client at that size is a multi-month project rather than one conversation. Grow everything around the account rather than cutting it. Fix the terms now so it stops getting worse: deposits, cancellation windows, payment terms in writing, a scope that says what’s included. Watch the share monthly the way you’d watch cash flow. Every point it comes down makes the decision less existential.

The calculation doesn’t decide whether the behavior is acceptable. It tells you whether how to fire a client is a conversation you can have now, or one you need to build the financial runway to have. Either way it belongs in your small business planning, not a folder marked difficult people.

How to fire a client without burning the bridge

When I do end one, it’s a conversation first and confirmed in writing after. Before the meeting, check the notice and termination terms in your own agreement and settle the final service date, unfinished work, and access transfer. Then call the decision-maker and tell them you don’t feel you’re the best partner for them.

Where the problem is fit, I try to have a referral ready, so I can say, “I don’t think we can service your company’s needs effectively, but I know this other great company who would be perfect for you.”

That referral is most of what separates a decent exit from a bad one, and it’s the part of how to fire a client that people skip. Having somewhere to send them is the difference between a rejection and a redirection. I do it outside of firings too: a lead I can’t or don’t want to service gets passed over with a warm introduction and no expectation of anything back.

Where the problem is conduct or chronic non-payment, no referral. You don’t hand that to somebody who trusts you. We vet incoming referrals like any other lead for the same reason: if it goes sideways it gets back to whoever introduced them.

I’ve lied before and said we’re too busy, but that feels icky. I’d rather be honest and tell them it isn’t working for us.

Where it’s a people problem, I’ll tell the contract signer how their team behaved with mine. They need to know how their people treat vendors, and then they can handle it. That isn’t a parting shot. It’s the only useful thing left to hand over.

Then the part nobody tells you about. If I don’t like working with a client enough to fire them, my team feels the same and we’ve already discussed it, so it’s never a surprise. When I tell them it’s done, it’s as much a weight off their shoulders as mine. It’s a moment for rejoicing and celebrating and new beginnings, and it breathes life into my employees. They work harder on the other accounts. They respect me as their leader for standing up for them.

I don’t fire clients often, so I’m no expert at this. But it’s a win-win all around, and I still get depressed every time. Both are true at once.

The version where you never have to do this

The best answer to how to fire a client is never needing to ask. Looking back at the ones I’ve let go, the tell was nearly always there at the pitch. Misaligned expectations are the biggest. If you sense a client will be unreasonable about timing or results, better to walk away and refer them than sign and have to end it later.

Easy to say when the pipeline is full and hard when it isn’t. I know that. We’re bootstrapped and we live and die on referrals, and I still haven’t built a system that makes them predictable.

But the arithmetic goes the other way from how it feels. Sign any business you can and you’ll burn bridges. There’s a capacity ceiling too: be realistic about how much you can service before your service level degrades, because stretch past it and the referrals dry up. Existing clients stop feeling the love they used to.

That’s the case for learning how to fire a client early rather than late. Not that it feels good, because it doesn’t. It’s that the client you should have turned down is costing you the ones you’d actually want.