A borrowed figure
is worse than none.
Search this question and you will be given a range, confidently, by someone who does not know which country you are in. The variables that move the answer are not details: rent in your city against rent two hundred kilometres away, whether you teach children in the afternoon or companies in the morning, whether teachers are employed or freelance, whether accreditation applies to you, and whether you open with premises at all. Any two of those can change the total by a multiple. A single figure that survives all of them does not exist.
A borrowed number is worse than no number because it feels like planning. Founders anchor to it, raise roughly that much, and then discover that their particular combination costs half again as much — usually in month five, when there is no longer a choice to make. The useful work is not finding the number. It is knowing which blocks the number contains, so that when your own estimate comes out uncomfortably large you can see exactly which decision is making it large and choose differently while it is still free to.
Build the estimate
from these.
Two notes on the second block, because staffing is where founders make the decision that quietly defines the business. Employing teachers gives you people who are available when you need them, who build relationships that keep students, and who can be asked to cover. It also means paying them in a thin August. Engaging freelancers converts that fixed cost into a variable one, which is genuinely safer while you are small — but you are then one of several clients rather than an employer, availability is not guaranteed in the week you need cover, and the arrangement is subject to employment rules that differ by country and are not optional. Whatever you choose, choose it knowing which risk you have taken rather than discovering it.
The third block resists cutting in a way the others do not. A school with no students does not need more space or more teachers; it needs to be found. Founders under pressure cut marketing first because it is the only line with no contract attached, and it is the line that produces the enrolments they are short of.
You are funding months,
not equipment.
Ask a founder what starting cost them and they list furniture, a website, a deposit. Ask what nearly closed them and it is always the same answer: the length of time between opening and the point where fees covered the month. That period has its own shape in language teaching. Enrolment is seasonal, so opening in the wrong month can cost you an entire cycle. First cohorts are small, because a school with no reputation converts fewer enquiries. And the students who do enrol often pay monthly rather than for a term, so even a healthy October arrives as twelve small payments rather than one useful one.
Work it out explicitly. Write down your monthly fixed costs once the doors are open. Estimate — pessimistically — how many students you will have in month one, month three and month six, and what each pays per month. Find the month where income crosses cost, add three months because it will be later than you think, and multiply the shortfall. That figure is a startup cost, as real as the deposit, and it belongs in the amount you raise rather than in a hope that it will be fine. A school that funds three months and needs eight does not fail because the idea was wrong; it fails with a full timetable and no cash.
Break-even is a class-size
question, not a student-count one.
The following uses round invented numbers purely to show the shape of the calculation. They are not market rates and you should replace every one of them with your own.
Say fixed costs are €4,000 a month — space, the person answering the phone, software, everything that does not vary with a student. Say a course sells for €200 a month per student, and a teacher costs €800 a month to run one class. A class of four brings in €800 and pays only for the teacher: it contributes nothing to the fixed costs and the school is losing €4,000 a month. A class of eight brings €1,600 and contributes €800 after the teacher. A class of twelve contributes €1,600. So break-even needs either five classes of twelve, or ten classes of eight, or an impossible number of classes of four.
The lesson survives the change of numbers: in a group-teaching business, survival is decided by average class size, not by total enrolment. Two hundred students spread thinly across thirty half-empty classes loses money; a hundred students in ten full ones does not. This is why timetabling is a financial function rather than an administrative one, and why a minimum viable cohort — the size below which a class does not run — is the most important number in the school. The timetabling article covers how to set and hold it, including what to do with the students in a class you had to cancel.
Administration is no longer
a capital purchase.
If you are budgeting from advice more than a few years old, one block is badly out of date. School administration software used to mean a licence, an installation and a consultant. For a school of the size described in this article it is now a subscription in the tens of euros, and the sensible approach while you are proving demand is to spend nothing at all: ours is free for one educator and ten students, €29 a month on Pro, and from €99 on Academy, with 0% taken from what students pay you. Whatever you choose, run the first cohort on a free tier and upgrade when the students arrive rather than before.
Two cautions so this block does not surprise you later. First, the subscription is not the whole cost — a platform that takes a percentage of tuition will eventually cost more than one that does not, and the comparison has to be run at the size you intend to be, which is the arithmetic in the total cost article. Second, this block does not include accounting. A school platform handles products, payments and what each family owes; it is not a bookkeeping or payroll system, and a school with staff will run one alongside it. Budget for both.