Everyone who has run a business for more than a couple of years has one. The client who looked like the making of you and turned out to be the thing quietly holding you back. Ours was on the books for four years. Two of those were good.
The uncomfortable part isn't that we took them on. Taking them on was a reasonable decision with the information we had. The uncomfortable part is the twenty-odd months where we knew, everyone in the business knew, and we carried on anyway.
They looked like security
They came in early, when the pipeline was thin and every new name felt like proof the whole thing might work. They were bigger than our other clients, they paid on time, and they were the logo we put at the top of the list when someone asked who we worked with. Within a year they were roughly a third of revenue.
That felt like stability. It was the opposite — but it takes a while to see the difference between a client who makes you secure and a client who makes you dependent, because both feel the same from the inside when the invoice clears.
What they actually cost
The drift was gradual, which is why nobody flagged it. A bit more scope each quarter, never enough to justify a difficult conversation. Meetings that ran long. A habit of sending work back on a Friday afternoon. Two people internally who were, unofficially, the people who dealt with them.
When we finally sat down and costed it honestly — actual hours, including the ones nobody bothered logging because they were 'just a call' — the margin on that account was a fraction of what we made on clients billing half as much. On some months it was negative. We had been treating a third of revenue as a third of the business, when it was closer to a tenth of the profit.
And that's only the part you can put in a spreadsheet. The rest was harder to quantify and probably cost more: the good work we didn't pitch for because capacity was committed, the two people who dreaded Thursday, the way an unreasonable request became normal because we'd absorbed the last five.
A bad client rarely costs you money in a way you notice. It costs you the better business you'd have built with that capacity.
A worked example
The figures below are illustrative rather than our actual accounts, but the shape of the answer is exactly right, and you can run the same sum over your own book in about twenty minutes. The only thing you need that most owners don't already have to hand is a fully-loaded hourly cost for each person — salary plus employer's National Insurance at 15% on earnings over the £5,000 secondary threshold, plus employer pension at the 3% auto-enrolment minimum, plus that person's share of premises, software and insurance. It is usually a third to a half more than the salary alone, which is precisely why costing a client at salary rates makes every account look profitable.
Say the big client bills £120,000 a year and the rest of the book bills £240,000. They are a third of a £360,000 turnover and, on the face of it, the best account you have.
Now cost the delivery. Two people spend an average of 28 hours a week between them on that account — 1,300 hours across a working year once holiday comes out. At a fully-loaded £42 an hour that's £54,600. Add your own time at four hours a week, 200 hours at £60, for another £12,000. Apportion overheads by headcount and add £18,000. Total cost of service: £84,600. Gross profit: £35,400, a margin of 29.5%.
Run the same exercise across the other £240,000 and suppose it costs £132,000 to deliver. That's £108,000 of gross profit at a 45% margin. Per pound of revenue, the big client returns about 65p for every £1 the rest of the book returns — which is the first uncomfortable number, and the one that stops the conversation being about how big they are.
Then add the line that never appears in any management account. Those 1,500 hours are capacity, not just cost. The rest of the book works out at roughly £110 an hour of billed time against the big client's £80. Refill even two-thirds of that capacity — 1,000 hours — with work at the same rate and margin as the rest of the book, and it produces £110,000 of revenue and about £49,500 of gross profit, against the £35,400 the account actually earns. On that arithmetic the client isn't merely less profitable. Keeping them costs roughly £14,000 a year.
That is the sum we hadn't done. Do it once a year, use a loaded hourly cost rather than salary, and include the hours nobody logs — the answer changes decisions in a way that a revenue table never does.
Why we didn't act
Fear, mostly, dressed up as commercial prudence. A third of revenue disappearing is a genuinely frightening prospect, and every month we didn't act, the number got a little scarier and the account a little more entrenched. There's a version of loss aversion specific to business owners: you'll tolerate a slow, certain bleed to avoid a fast, uncertain one.
There was ego in it too. Losing a client feels like a verdict on you, even when you know perfectly well it's a mismatch rather than a failure. And there was a genuine, decent instinct — these were people we'd worked with for years, and nobody wants to be the one who walks away.
None of those reasons survive contact with the actual numbers. All of them survived indefinitely while we avoided looking at the actual numbers. That's the real lesson: this isn't a courage problem, it's a measurement problem. We simply didn't have per-client profitability in front of us often enough for the truth to become unavoidable.
How it finally ended
Not dramatically, in the end. We priced the renewal at what the work was genuinely worth — a significant increase, with a scope written down properly for the first time in three years, and a clear line about what sat outside it. We were straightforward about why: the work had grown, the price hadn't, and we couldn't keep doing it at that number.
They said no. We'd braced for a row and got a fairly civil parting. In hindsight the increase was a way of asking a question we'd been avoiding — is this relationship worth what it costs both of us — and both sides already knew the answer.
The interesting part is that they weren't villains. They were a client behaving exactly as a client will behave when nobody has ever told them where the edges are. We'd trained them to expect it, one absorbed request at a time.
What replaced it
The gap was real and it hurt for a quarter. What surprised us was how quickly it filled — not with one big replacement, but with three smaller clients we'd previously have said we had no capacity for. Total revenue took a few months to recover. Profit recovered faster than revenue, because the new work was priced properly from the start and didn't come with an invisible tail of unbilled hours.
The bigger change was internal. The team got noticeably better within weeks, which told us something about what that account had been costing in energy that no P&L was ever going to show.
What we'd do differently
Review profitability per client, not just revenue per client, at least twice a year — it's the single habit that would have caught this two years earlier, and it belongs alongside the other numbers every owner should know. Set a concentration limit and treat crossing it as a signal to go and win more work, not as an achievement. Write down scope properly at the start, and reprice annually as a matter of course rather than as a confrontation.
And when something has felt wrong for six months and the numbers agree with the feeling, act on it. The version of this story where we'd moved earlier isn't a story at all. It's just a slightly better business, two years sooner.
Common questions
How do I actually end a client relationship without it turning nasty?
Give the notice your contract requires, in writing, and offer a proper handover — that combination defuses most of the heat. Read the agreement first: many service contracts carry 30, 60 or 90 days' notice, and walking away sooner is a breach you can be sued over even when the client is the difficult one. Say what is ending and when, not why they were hard work; a short, factual letter ages far better than a candid one. Then finish the work in the notice period to the standard you'd want quoted back to you, hand over files and passwords, and invoice normally. If you would rather not sack them outright, repricing the renewal to what the work is genuinely worth lets them make the decision instead.
How much of my turnover is it safe for one client to be?
There is no legal limit, so the useful test is a survival test rather than a percentage. Take your biggest client's monthly fee, assume they give notice tomorrow, and work out how many months your cash and pipeline cover the hole — if the answer is under three, that account is a concentration risk whatever share of revenue it represents. The number matters commercially too: lenders reviewing a loan application and buyers valuing your business both discount heavily for a book that leans on one name, and a heavy concentration can knock a chunk off a sale price or turn a facility down outright. Set your own ceiling, write it down, and treat crossing it as a signal to go and win more work.
They owe me money and I want to resign the account. What can I claim?
You can charge statutory interest and a fixed recovery sum on every overdue business-to-business invoice, whether or not your contract mentions it. The Late Payment of Commercial Debts (Interest) Act 1998 sets interest at 8% above the Bank of England base rate — base rate has been 3.75% since it was held on 30 July 2026, giving 11.75% — fixed by reference to the rate in force on the preceding 30 June or 31 December. On top of that you can claim a fixed sum for recovery costs: £40 for a debt under £1,000, £70 up to £9,999.99, and £100 for £10,000 or more. You normally have six years to sue on the debt, and the Small Business Commissioner will look at complaints about late payment by larger customers.
How do I work out profit per client if my bookkeeping doesn't track it?
Estimate it once, roughly, rather than waiting for a system that tracks it perfectly. Take your four or five biggest clients, and for a normal month ask everyone who touches those accounts for an honest hours figure including calls, emails and the rework nobody logs. Multiply by a fully-loaded hourly cost — salary plus employer's National Insurance at 15% above the £5,000 secondary threshold, plus the 3% pension minimum, plus a share of premises and software — not by salary alone. Add an overhead share by headcount. Compare the result to what each client actually pays. A twenty-minute estimate that is 20% out will still tell you which accounts are underwater, which is the decision you are trying to make.
Won't putting the price up just lose them?
Sometimes, and that is the point of doing it — a repricing is a way of asking whether the relationship is worth what it costs both sides. Price the renewal at what the work genuinely costs to deliver plus the margin the rest of your book earns, write the scope down properly, and say plainly what sits outside it. Give a reason that is about the work rather than about them: the scope has grown, the price hasn't, and it can't continue at that number. Some clients accept, and you have fixed the account without losing it. Some decline, and you have recovered capacity you can sell at a better margin. Both outcomes beat another two years of absorbing it quietly.
