Break-Even Is a Feeling

I knew the number to the unit. Five hundred.

Five hundred units a month and the business washed its face. I had it written on a card, taped inside a cupboard door where I'd see it every morning, and for about seven months I ran my entire life against that figure — good week, bad week, ahead, behind — with the specific comfort that comes from having a target you can count toward.

Then, in a completely unrelated conversation with an accountant, I mentioned my variable cost per unit and she asked whether that included the payment processing, and it did not.

Three pounds a unit. That was the whole error. And my break-even wasn't five hundred.

It was five hundred and eighty-nine, which is not a small revision when you've spent seven months believing that the difference between those two numbers was the difference between a business and a hobby.

The formula, and what it's made of¶

Break-even units = fixed costs ÷ contribution margin per unit, where contribution margin is price minus variable cost.

It's a good formula. It's the right question. And it has a property nobody mentions when they teach it: it is a ratio of two estimates, and ratios amplify error in the denominator.

Take a business with £10,000 a month in fixed costs, selling at £50 a unit, with variable costs it believes are £30.

Now read the error transmission, which is the point of the table.

Getting the variable cost wrong by £3 — a 10% error — moves break-even from 500 to 589 units. That's a 17.6% error in the output. Getting it wrong by £5, a 16.7% error, moves break-even to 667: a 33.3% error.

The formula roughly doubles whatever you got wrong.

That's the false precision. The output arrives with the authority of arithmetic, carrying error from inputs that were, in my case, a number I'd worked out once on a Sunday and never revisited.

And it compounds with something this desk has covered from the other direction: estimates are not randomly wrong, they're wrong in a consistent direction. If your variable-cost guess is systematically low — and guesses about your own costs almost always are, because you forget the small recurring ones — then your break-even is systematically optimistic, every single time you compute it.

It isn't a point¶

Here's the more important error, and I made it for years.

I thought of break-even as a threshold. A line you get over, once, after which you are a different category of business. That's how it's discussed — we hit break-even in March — and it's how it felt, taped inside the cupboard door.

But it's a rate. Five hundred and eighty-nine units a month, every month, and a month you miss is not repaid by a month you exceeded. The fixed costs arrive again on the first regardless of what happened in the previous thirty days, because that is what a fixed cost is: a decision made once that keeps executing.

That distinction has a direct consequence for how much cushion you need. If break-even were a threshold, you'd need enough cash to reach it once. Because it's a rate, you need enough to survive the months you're under it — which is a completely different question, and the one buffer days actually answers.

The BLS keeps the tally of how that goes. Its Business Employment Dynamics series tracks private-sector establishments from opening, and for the cohort that opened in March 1994:

Read that as a break-even curve rather than a failure curve. Almost four in five clear the first year. Half are still there at five. The attrition isn't concentrated at the start — it accumulates steadily, which is what you'd expect from a condition that has to be met repeatedly rather than a hurdle cleared once, and it's a useful corrective to the way failure statistics usually get quoted.

The turn¶

So why does it feel like a point?

Because the feeling has a different trigger than the arithmetic, and the trigger is usually a single good month.

You have a strong month. Money in the account, invoices paid, the number on the card cleared. And what registers is not the rate has changed but we made it — a transition rather than a data point. I remember the month I first cleared five hundred units and I remember the specific physical sensation of it, and I do not remember any of the four subsequent months in which I didn't.

That asymmetry is the expensive part. The good month is experienced as a change in identity; the bad months are experienced as noise. So the emotional break-even arrives early, well before the rate is sustainably met, and it arrives attached to a decision — hire someone, sign the longer lease, take the bigger space — made on the strength of a category change that hasn't happened.

And when the arithmetic is also optimistic, because the variable cost was a guess made on a Sunday, the two errors point the same way. That's the combination that gets people: a formula biased low, and a psychology that fires on the first month that clears it.

The honest version of the cupboard card would have said: 589, on average, sustained, and if we do it in March that means nothing about April.

That's a considerably worse thing to look at every morning. It's also the true one.

What to do instead¶

1. 1

   Rebuild the variable cost from bank lines, not from memory

   Take one month of actual outgoings and assign every line either to fixed or to per-unit. The per-unit column is always longer than the version in your head — payment fees, packaging, returns, the delivery you absorbed. This single exercise moved my number by 17.6% and I had been running on the old one for seven months.

2. 2

   Compute three break-evens, not one

   Your best guess, and then the same calculation with variable cost 10% and 20% higher. You now have a range instead of a false point, and the range is the honest output. Put the pessimistic one on the card.

3. 3

   State it as a rate, out loud, with a period attached

   "589 units a month, sustained" rather than "589 units." The two extra words do most of the psychological work, because they make a single good month legible as one observation rather than as an arrival.

4. 4

   Track a rolling three-month average, not the month

   Monthly figures are too noisy to feel accurately, which is exactly why one good one gets over-read. A three-month rolling average moves slowly enough that when it crosses, something has actually changed. This is the cheapest available defence against your own good news.

5. 5

   Recompute whenever a price or a supplier moves

   Break-even is not a constant. Every price change, every supplier increase, every new payment processor silently moves the denominator — and because the formula amplifies, a small change in input is a real change in the number taped to your cupboard. If you're revisiting it anyway, revisit what you're charging at the same time, since price and margin are the same lever seen from two ends.

The card¶

I replaced it eventually. The new one said 589, and underneath, in smaller writing, per month, average of the last three.

It was a considerably less satisfying object. You cannot feel triumphant about a rolling average — it's designed specifically to be unmoved by the thing that would otherwise move you, which is the entire point and also why nobody makes posters of it.

What it did was end a particular kind of month. The month where a good four weeks produced a decision that a good four weeks did not support: the hire, the space, the commitment made on the strength of a feeling that had arrived about a hundred units early and been treated as news.

The number was never the problem. The problem was that it was a guess wearing the clothes of a measurement, and I had taped it somewhere I would see it every morning, which is a very effective way to stop asking whether it was right.

Rebuild it from the bank statement. Compute it three times. Then write down the worst one.

It's just business.