A 20% rise survives losing one client in six

You can lose considerably more clients than you think. For someone selling their own time, the break-even is c = p ÷ (1 + p): a 10% rise survives losing 9.1% of your clients, a 20% rise survives 16.7%, a 50% rise survives 33.3%. Below those thresholds you are ahead. And because a departing client hands back delivery capacity, the profit break-even is more forgiving still.

The fear is unbounded. The number is small. That asymmetry is most of what is wrong with pricing in expert practices.

The break-even, before the argument for it

Share of clients you can lose and still earn the same, when marginal delivery cost is close to zero. c = p ÷ (1 + p).

Price increase	Clients you can lose and break even
+10%	9.1%
+20%	16.7%
+30%	23.1%
+50%	33.3%
+100%	50.0%

16.7% of your clients can leave after a 20% price rise before the rise costs you anything — and every one who goes hands back the hours you were spending on them.

Ten clients and a 20% rise: the break-even is 1.67 clients. Clients do not come in fractions, so you can lose one and be ahead — nine at 120% of the old price is 10.8 units of revenue against 10. Lose two and you are down 4%.

One client. That is the size of the thing people spend two years not doing.

Where the formula comes from

It is worth deriving rather than asserting, because the derivation tells you when it applies and when it does not.

Let the current price be P and the number of clients N. The variable cost of delivering to one client is V, so contribution per client is M = P − V, and contribution margin as a fraction of price is m = M ÷ P.

Total contribution today is N × M.

Raise the price by a fraction p. The new price is P(1 + p). Variable cost is unchanged, so contribution per client becomes P(1 + p) − V = M + pP. Lose a fraction c of clients and you keep N(1 − c) of them.

New total contribution: N(1 − c)(M + pP).

Set the two equal, which is what break-even means:

N(1 − c)(M + pP) = N × M
1 − c = M ÷ (M + pP)
c = pP ÷ (M + pP)

Divide top and bottom by P and it resolves to c = p ÷ (p + m). That is the general form and it holds for anything sold at a margin.

Now the case in front of us. A solo expert selling hours they were going to work anyway has a marginal delivery cost close to zero — your time is already committed and already in the fixed base. So m is close to 1, and the formula collapses to c = p ÷ (1 + p).

Which is the harshest version of the formula there is, and that is worth dwelling on. As margin falls, the break-even rises. A retailer on a 30% gross margin, m = 0.3, taking a 20% price rise: c = 0.2 ÷ 0.5 = 40.0%. They can lose two-fifths of their customers and be no worse off. You, at m close to 1, can only lose 16.7%.

Lower margin, more forgiving. The formula is at its least generous exactly where you are standing, because you have no variable cost to save when a client leaves. And it is still 16.7%.

The half of the calculation nobody does

Here is the part the standard treatment leaves out, and it is the part that changes the decision.

Ten retained clients at CHF 2,000 a month is CHF 20,000. Raise to CHF 2,400 — a 20% rise. Two of the ten leave. That is a 20% loss, worse than the 16.7% break-even, so on the revenue arithmetic you have failed: 8 × CHF 2,400 = CHF 19,200, down CHF 800, a fall of 4%.

Except you have just recovered a fifth of your delivery capacity. Two clients' worth of hours came back, and they came back permanently.

Win one replacement at the new price and the book is 9 × CHF 2,400 = CHF 21,600. That is CHF 1,600 more than you started with — 8% more money from nine clients instead of ten, which is 10% less delivery.

You would have to lose one client in six before a 20% rise cost you anything at all — and you would have bought back the better part of a day a week in the process.

The rate you defend at the door and abandon inside the house

The Freelancer-Kompass 2026, freelancermap's survey of 5,412 freelancers across the DACH region fielded between November 2025 and February 2026, contains two findings that belong next to each other.

70% of respondents reject new projects because the hourly rate on offer is insufficient. And 62% plan to keep their rates unchanged, against 29% planning increases and 9% planning decreases.

So the reservation price exists. It is real, it is firmly held, and it is enforced — on strangers, at the point of enquiry, with no apparent difficulty. It is simply never applied to anybody already on the book.

That asymmetry has a shape worth naming. The price is a boundary at the door and a memory inside the house. Which means the longest-standing clients — the ones who have received the most of your improvement, who bought you when you were least experienced, who are least likely to leave because switching would cost them the most — are the ones paying the least. The book is priced in reverse order of loyalty and expertise.

It is also, as behaviours go, unusually addressable. Nobody in that 70% lacks the capacity to say a number out loud. They have demonstrated it repeatedly.

Two continents, two professions, the same 62%

Heard's 2026 Financial State of Private Practice Report surveyed 1,950 private-practice therapists across all fifty US states and the District of Columbia, on 2025 data. 62% have no plans to adjust their fees in 2026. Only 33.1% raised fees in 2025.

Sixty-two per cent among DACH freelancers, mostly in IT and consulting. Sixty-two per cent among American therapists. Two independent surveys, two unrelated professions, two continents, the same figure to the percentage point.

Set against it: Consulting Success, surveying roughly 1,000 consultants across more than 75 countries, found 79% actively looking to increase their fees. The same study found 39% have never used value-based pricing because they do not know how, and 25% will lower their fees to win a client.

These are different samples, so the numbers do not chain arithmetically. But 79% wanting and 62% not moving is hard to explain as a knowledge gap. The gap between intending and doing is the whole of the problem, and it is not solved by learning another pricing method.

Heard also reports what happened to the third who did move. Therapists who raised fees in 2025 reported median revenue of $94,792, against $74,979 for those who did not — a difference of $19,813, or 26.4%.

That is correlation and not established causation, and I would rather say so than let you read it as a promise. A practice with a waiting list is both more able to raise its fee and more likely to earn more anyway; the fee rise may be a symptom of demand rather than a cause of income. What the figure does establish is narrower and still useful: the group that raised its fees did not end up poorer. Collapse is the outcome the 62% are protecting against, and in that sample it did not appear.

Why "add a few per cent and communicate early" fails

Simon-Kucher's Global Pricing and Sales Study, covering roughly 2,000 companies, found that 97 out of 100 failed to achieve their price-increase targets, realising only 32% of what they planned. A 5% target delivered about 1.6%.

Set against inflation, a 1.6% increase in a year of 2% inflation is (1.016 ÷ 1.02) − 1 = −0.4%. The price increase was a pay cut with extra meetings.

The mechanism is the interesting part, and it transfers to a one-person practice in an uncomfortable way. In a company, the increase is decided centrally and conceded locally — head office sets 5%, and the person facing the customer gives most of it away, because they are the one absorbing the discomfort. A solo practice compresses both roles into one person. Which removes the only structural defence that exists in the corporate version: somebody else set the number, and you are not authorised to change it.

That study is from 2017 and covers companies with sales organisations rather than expert practices. The setting does not transfer. The mechanism does, and in the solo case it is worse rather than better.

What holding a rate costs while you decide

Holding a price is not a neutral act. It is a decision to take a pay cut in instalments, and the instalments are invisible because the number on the invoice does not change.

The arithmetic is standard: real change = (1 + nominal change) ÷ (1 + inflation) − 1, compounded over the years held. Take a €120 rate held flat for three years.

Real value of a €120 hourly rate held unchanged for three years, at three illustrative inflation rates. Rates and inflation figures are illustrative, not survey data.

Annual inflation	Real value after three years	Effective pay cut
2.0%	€113.1	5.8%
2.5%	€111.4	7.1%
4.0%	€106.7	11.1%

And that is the flat case. freelancermap reports the average DACH freelance rate at €103 for 2026 against €104 for 2025 — the first nominal decline since the study began. A nominal fall on top of inflation is not a pause. It is a reduction taken twice.

What this means in practice

The break-even is a threshold, not a forecast. It tells you what you could survive, not what will happen. That distinction matters because it changes what the number is for: it is not a prediction to be argued with, it is a floor under a decision that currently has none.

Three things determine whether the general formula describes your situation.

The first is concentration. If two clients are 60% of your revenue, 16.7% is not a meaningful average — the relevant question is whether either of those two would leave, and that is a question about two specific relationships. Average break-evens are for books that look like averages.

The second is who leaves. The formula assumes a random draw, and the draw is never random. The clients most likely to go over price are the most price-sensitive, which is generally to say the least profitable and the most demanding. In practice this makes the real outcome better than the formula predicts, which is one of the few places where the conservative assumption runs in your favour.

The third is replacement. A 20% rise plus one new client at the new price was +8% on 10% less delivery in the example above. Whether you can win that replacement is not a pricing question at all — it is a question about whether you have a pipeline that runs while you are busy. That is why price and pipeline are the same conversation, and why raising a price with an empty pipeline is a different act from raising one with a full one.

There is also the sequencing question. If your capture rate is a third — if only a third of your working hours turn into collected money — then a 20% rise is a 20% improvement on a third. The four terms behind that fraction are usually the larger prize, and the question of whether the rate is genuinely low has its own three numbers behind it.

What this cannot tell you from the outside

c = p ÷ (p + m) is arithmetic. It cannot tell you what will happen, because what happens depends on facts about your particular book that no survey contains: how concentrated it is, which clients hold contracts and which hold habits, what it would cost each of them to replace you, and whether your price is being compared to a competitor or to the cost of the problem you solve. Those four facts decide the outcome, and none of them is in any of the studies above.

They are, however, all knowable. A concentration table takes an afternoon. A read on which clients are price-taking and which are price-setting takes a conversation with yourself about each name on the list. The break-even applied to your own book rather than to an average takes ten minutes once the first two exist.

What is telling is that nobody has been engaged to produce them. Your accountant records the invoices you raised, never the ones you could have raised. Your lawyer drafts what has been agreed. The concentration table, the price-sensitivity read and the break-even on your actual book are not anybody's job, which is why a decision worth a fifth of your income gets made on a feeling about one client's face.