Nobody has spare hours. A study plan that adds time to a week is a wish with a timetable attached, and it survives about three weeks, which is exactly as long as the enthusiasm that wrote it. A plan that works is a subtraction: it names the hours that already exist, decides which of them are being converted, and protects them against the things that will come for them.
The second design rule is less obvious and it matters more. A plan has to be built for the worst week rather than the best one, because the bad week is the one that decides whether the plan still exists in month three.
The six that make a plan survive
- daily beats long. Twenty minutes a day outperforms three hours on Sunday, because memory responds to frequency.
- fix the time, not the length. A slot in the day is kept; an intention to study is not.
- plan the week, not the year. Long plans are abandoned at the first disruption and rebuilt from nothing.
- one course, finished. Starting four and abandoning each is the single commonest pattern among people who stall.
- measure hours, not intentions, and write them down, because the record is what shows a plateau is a slowdown rather than a stop.
- build in a rest day. A plan with no slack breaks permanently the first time life interferes.
Do the subtraction before writing anything down
A week has 168 hours. Take out 56 for sleep, 50 for work and the travelling around it, 40 for family, meals and everything administrative, and 15 for whatever else is already fixed in your week, and 7 are left. That last figure is the entire budget, and it is the number a plan has to be written against rather than the number of hours the plan would like.
Run those subtractions on your own week rather than reading mine, because the interesting part is where yours differs. Two things usually come out of it. The budget is smaller than the plan people had already written, often by a factor of three. And a good share of the hours removed in the middle rows are not actually fixed: they contain audio, or a queue, or a commute, and those can carry the language without being given up. Which hours those are, and what they can hold, is filling your day with the language without moving country.
Book the rarest hour first
The three kinds of study hour do not substitute for one another, and they also do not need the same amount of room. Retrieval is short, dull and daily. Material is long and flexible and comes out of hours you were already spending. Time with another person is rare, costs money, and is the only one of the three with a deadline attached to it.
That last property is why it goes into the calendar first, before anything else. A lesson at seven on Thursday is an appointment with a person who will notice, and it is the only item in a study week that cannot quietly be moved to tomorrow. Everything else arranges itself around it: the deck gets done because Thursday is coming, and the material gets chosen because it has to be worth bringing. Take the booked hour out and the whole week becomes negotiable at once, which is the real reason a plan built entirely out of solo work fails, and it has nothing to do with discipline. Choosing and keeping that hour is finding someone to speak to, and what to do in the hour.
The weekend catch-up eats the weekdays
This is the stall, and it has a mechanism rather than a moral. The plan says twenty minutes on five weekdays. A Tuesday goes wrong, so it moves to Saturday. Then a Wednesday goes wrong. Saturday is now an hour, then two, and the weekend session becomes the place where the week gets repaired.
What has happened is that a daily habit has been converted into a weekly one, and no single day announced it. A weekly session is worse in two separate ways: the spacing is wrong for how memory actually works, and a habit that fires once a week never becomes automatic, so it costs a decision every time. The exit is to refuse the repair. A missed day is written off, not moved, and the plan needs a floor low enough to make that rare: ten minutes on the worst day of your worst week, small enough that skipping it is never justified. That floor, rather than any ceiling, is what streaks, restarts and what to do after a bad month turns on.
Three kinds of hour
How much room each kind needs, and which one goes in the calendar first
The claim is volume rather than order. The narrowest tier is the one to book first, because it is the only appointment in the week.
- An hour with a personRare, paid for, and the only one somebody will notice you missing
- Material you can nearly followLong and flexible, converted out of hours already being spent
- Ten minutes of retrieval, dailyShort, dull, non-negotiable, and holding up both tiers above it
The tiers do not substitute in either direction. Ten hours of material will not produce what one booked hour produces, and a booked hour with no base under it is sixty minutes of being corrected on things that will be gone by Friday.
Two things to set up, and neither is an app about learning
Put the sessions in your ordinary calendar as events with a start and an end, in the same calendar as your work. It replaced the habit tracker, which records whether you did it and never defends the time, and the difference is that an event is visible to the part of your week that schedules meetings over you. A named block at seven is declined by other things. A private intention is not. Free, and it takes ten minutes once.
The second is a timer, and Toggl Track in 2026 is still the least annoying one with a free tier that covers this. Start it when the session starts, stop it when you stop. It replaced estimating your own study hours afterwards, which everybody does and everybody gets wrong in the same direction. After a fortnight you have the real number, and the real number is what makes hours to fluency, by language and by starting point a calculation you can actually do instead of a claim you read.
The mistake: planning by material instead of by time
A plan that says finish this book, complete this app's course, watch this series is not a plan, because none of those has a rate attached to it. Material is unbounded, arrives faster than it can be used, and a list of it is a shopping list rather than a schedule. Worse, it hides the failure: falling behind on a book feels like a reason to buy a different book.
Write the plan in hours and kinds of hour, and let the material be whatever fits into them. Ten minutes of retrieval daily, forty minutes of material on four days, one booked hour a week. That is a plan you can grade at the end of the month with a stopwatch rather than a feeling, and it is what seven days built around the time you actually have is for. It also has the property that a bad month is visible as a number instead of as an atmosphere.
What a plan cannot decide
Hours are the input and they are not the whole answer, because the same seven hours produce very different results depending on which language they are aimed at and what you already speak. That difference is large, it is published, and it is the subject of difficulty, usefulness and the honest reasons people pick. If the calendar is full and correct and the language has stopped moving anyway, the cause is not scheduling and the diagnosis is in the stalls every learner hits, and the way through each.