Why students quit, and what keeps them

Teachers tend to interpret a student leaving as a verdict on the teaching. Usually it is not. Students leave at three predictable moments, for reasons that have more to do with the shape of language learning than with any particular lesson.

Knowing where the cliffs are makes them largely avoidable.

The three cliffs

Lessons
  1  2  3        8  10  12  14        20
  ●──●──●────────●───●───●───●─────────●
     ↑                ↑              ↑
   CLIFF 1         CLIFF 2        CLIFF 3
   "not what      the month-3     any 2-week
    I expected"    plateau          break

Cliff 1 — lessons two to three

The student expected something different. Usually one of: more conversation, less conversation, more structure, or less correction.

This is almost entirely preventable, and the prevention happens in the first lesson by stating explicitly how lessons will run and asking what they want. Then asking again at lesson three, when they know enough to have an opinion.

Two questions, ninety seconds:

“Is this what you expected? What would you change?”

Most students will not volunteer dissatisfaction. They will answer a direct question, and they will simply stop booking if never asked.

Cliff 2 — month three, the plateau

The big one, and the one that is not about you at all.

Beginner progress is enormous and obvious. From nothing to ordering food, introducing yourself, handling a shop — every week produces something visibly new.

Then it stops feeling that way. The gains continue but become invisible: vocabulary broadening, structures consolidating, listening speeding up. None of it produces a moment where the student notices a new capability.

From inside, this is indistinguishable from having stopped improving. And a student who believes they have stopped improving stops paying.

The intervention is evidence, not encouragement. “You’re doing really well” is not persuasive to someone whose direct experience says otherwise. What works is comparison against their own past:

• Play them a recording from three months ago. This is the single most effective retention tool available and almost nobody does it. It requires having recorded them in month one, which requires deciding to, once.
• Count something. “In September you could produce about forty sentences of this kind. You now have two hundred.” Numbers beat impressions.
• Restate the original goal and measure against it. They said they wanted to handle a work call. Can they now? Partly? That partial is the progress.
• Name the plateau in advance. Told in lesson one that month three feels like standing still, a student who reaches month three and feels that way interprets it as expected rather than as failure. This costs one sentence and works remarkably well.

Cliff 3 — the break

A holiday, a work crisis, illness. Two weeks off becomes three, then the student feels behind, then embarrassed, then it is easier not to come back.

Break recovery is a solved problem and the solution is a message:

“No rush at all — whenever you’re ready. We were in the middle of the past tense and your /r/ was starting to come. Want to pick up there?”

Specificity is what makes this work. A generic “hope to see you soon” reads as marketing. Naming what they were working on reads as someone who remembers them, and removes the fear that they will have to start over.

The underlying failure: invisible progress

Every cliff except the first comes down to the same thing. Students quit when they cannot see progress, not when they stop making it.

Which makes progress visibility a core part of the job rather than a nicety. Three things make it visible:

Recordings over time. Record two minutes in lesson one. Record two minutes every couple of months. This is the strongest evidence that exists and it costs nothing.

An accumulating artefact. Lesson notes that build into something with visible size — a deck, a document, a list of sentences that grows. A student who can see 300 sentences they now know has a different relationship to the plateau than one who has three loose Google Docs. See lesson notes students reuse.

Goals restated and checked. The goal from the first lesson, revisited at month two and month four, with an honest assessment against it. Not “are you happy” but “can you do the thing you came here to do, yet”.

Things that help more than they look like they should

Start every lesson from the last one’s notes. Two minutes. It makes the previous lesson feel like it counted, and it makes homework non-optional through washback rather than through nagging.

Book the next lesson before the current one ends. A student with a booked slot returns. A student who has to decide to book has to decide every week, and the drift compounds.

Send notes within the hour. Value delivered after the lesson extends it. Value delivered two days later arrives after the material has decayed.

Small wins engineered into every lesson. One thing they could not do at the start and can at the end. Not implicit — say it: “Two months ago you couldn’t have said that sentence.”

Ask, at month two and month five. “Is this working for you?” Students almost never volunteer dissatisfaction and almost always answer a direct question.

Things that help less than teachers think

More content per lesson. Cramming does not increase perceived value; it usually decreases it, because nothing is retained and the student feels they are falling behind their own lessons.

Elaborate materials. A beautiful slide deck does not affect retention. What the student can do after the lesson does.

Discounts. A student leaving because progress is invisible does not stay for 10% off. Price is rarely the actual reason, and discounting trains the wrong response.

Being nice. Necessary, insufficient, and frequently mistaken for sufficient. Warmth without visible progress buys three or four extra lessons.

The message to lapsed students

Send it once, at about three weeks of silence. Short, specific, no pressure, no discount:

“Hi — no pressure at all, just wanted to say the door’s open whenever. We’d got to the point where your past tense was almost automatic, and it would be a shame to lose that. Happy to pick up exactly where we left off.”

A meaningful fraction come back, because most lapsed students did not decide to stop. They drifted, and then the awkwardness of returning grew larger than the awkwardness of not. Removing the awkwardness is most of the work.