Running a school
Why students leave a language school
Published 17 August 2026
A school usually finds out that a student has gone when a payment does not arrive. The decision is weeks older than that, and it was made on evidence the timetable had already recorded.
Nobody leaves in the month they stop paying
A student who cancels in November decided in October and started deciding in September. What arrives at the end is the paperwork: the message that the schedule no longer works, the invoice left unpaid, the seat that stays empty. By then there is nothing left to save.
This is why retention gets treated as luck. A school that sees only the last step sees an event with no cause, and an event with no cause cannot be worked on. The causes were visible for weeks - they were simply never anybody's job to look at.
Four reasons, and only one of them is the teaching
Students leave for a small number of reasons, and schools reliably guess the wrong one. In the order they actually decide it:
- Progress stopped being visible. The student is still improving, but nothing shows it. No score, no level moving, no piece of work returned with a mark on it - so the honest feeling is that eight months have bought a vague sense of being slightly better.
- The rhythm broke. One lesson moved, then another, then the student missed one and nobody put it back. Attendance is a habit before it is a decision, and a habit survives about two disruptions.
- They fell behind and could not say so. One missed unit becomes two, the group moves on, and asking for help in front of five people is worse than quietly not coming. Nobody gives this reason out loud; it is reported as “no time”.
- The teaching genuinely did not fit. Wrong level, wrong pace, wrong purpose. It is the real one, it is the rarest, and it is the only one most schools ever ask about.
What leaving looks like before it is announced
None of this is invisible. Every reason above leaves a mark in the record a school already keeps, usually weeks before any notice:
- Homework comes back later each time, and then stops coming back. This is the earliest signal there is, and it moves first.
- A student starts missing one particular slot - the late one, the one after work - rather than lessons in general.
- The questions stop. Someone who has decided to leave stops asking things, because the answers are no longer for them.
- An invoice is paid late for the first time in eight months.
- A lesson is moved and never rebooked. It was not rescheduled; it was dropped, politely.
A school watching those signals has three to six weeks of warning. A school watching the bank account has none, because a bank account reports a decision instead of a doubt.
The weeks that decide it
Two moments carry most of the risk, and neither is the one a school worries about.
The first month. Attendance is not a habit yet and every lesson is still being judged. A student who misses one of their first four lessons is far likelier to leave than one who misses one of their next twenty - the same absence, at a different point, means something else entirely.
The plateau. Somewhere in the middle, progress stops feeling like progress: the distance between one level and the next is longer than the distance between the two before it, and the student reads the flat stretch as having stopped learning. Nothing has gone wrong with the teaching. What is missing is evidence.
What a longer stay is worth
Take the same group as the first piece in this series: six seats at €12, one ninety-minute lesson a week, a teacher at €22 an hour, €60 to win each student. Change nothing except how long people stay.
| Average stay | Lessons attended | Acquisition per lesson | What the hour leaves |
|---|---|---|---|
| 4 months | 16 | −€22.50 | −€2.50 |
| 8 months | 32 | −€11.25 | €8.75 |
| 12 months | 48 | −€7.50 | €12.50 |
| 18 months | 72 | −€5.00 | €15.00 |
The same teacher, the same price and the same room. Only the average stay changes. At four months the group does not break even; at eighteen it earns nearly twice what it earns at eight.
Per student it is the same story: sixteen lessons at €12 is €192 of lifetime revenue, forty-eight is €576. Three times the income from the same person, for the same €60 it cost to find them.
What to change first
- Give progress somewhere to show. A level that moves, a score that accumulates, work returned with a mark on it. The point is not measurement; it is that a student who cannot see progress will eventually conclude there is none.
- Defend the first four lessons. Call after the first absence rather than the third, and move the lesson instead of letting it lapse. It costs one message, and it is the highest-return message a school sends.
- Make falling behind survivable in private. A student who can catch up a missed unit on their own, before the next lesson, comes to the next lesson.
- Watch homework, not revenue. It moves weeks earlier, and it is already in front of you.
Where the signals live
Every one of those signals is already in the record: who handed in and when, who is missing which slot, which invoice went out late, which student has not asked anything in a month. They are invisible only while they are spread over a spreadsheet, a calendar and a chat thread. That is what EdTechCore puts in one place - the lesson on a shared board, homework and grading on one path, and the school's books behind them - so the first thing a school learns about a student leaving is not the payment that did not arrive.