Quick answer
Quick answer: Train in sessions short enough that you finish wanting more. Fifteen minutes daily for a month beats three hours on two Sundays, and it isn't close. The failure is almost never laziness, it's a schedule nobody could keep.
The second failure is training while angry after a loss, which teaches your brain that chess feels bad.
I read the research summaries on spacing before writing this, because the interval between sessions does more work than the sessions do.
Why week three is where people quit
The pattern is consistent enough to plan around.
Week one is exciting. Week two is fine. Week three is when the early gains stop and the rating stops moving, and the session starts feeling like homework.
Nothing has gone wrong at that point. That's just what the curve looks like.
But the schedule people set in week one was built for week one enthusiasm, and it collapses the moment the enthusiasm does.
Knowing it's coming is most of the defence. If you expect week three to feel flat, it stops being evidence that you've hit your limit.
Build the schedule for your worst day
This is the single most useful change.
Don't ask what you can do on a good day. Ask what you could do on a day where you're tired, busy and not in the mood.
Whatever that number is, that's your daily amount. For most people it's ten or fifteen minutes.
A plan you keep on bad days beats an ambitious plan you keep for nineteen days and then abandon.
The same logic applies to the time of day. Pick the slot that exists on a bad day, not the one that exists on a free Saturday.
Why spacing does the work
Memory research is unusually clear on this. Material reviewed at intervals sticks far better than the same material crammed.
The technique built on it, spaced repetition, shows difficult items more often and easy ones less often.
Tactics work the same way, because a tactical pattern is a memory item with a shape attached.
So the gap between sessions isn't wasted time. It's part of the mechanism, and cramming removes it.
Stop while you're still good at it
There's a point in every session where your accuracy drops and you start guessing.
Everything after that point is worse than useless, because you're rehearsing sloppy looking rather than careful calculation.
So the rule is to stop at the first sign of guessing, even if that's after eight minutes.
Ending a session while it's still going well also means you look forward to the next one, which is the whole game.
The hardest part is that the guessing point moves. On a tired day it arrives after five minutes, and the honest response is to stop after five minutes.
The tilt problem
Losing a game you should have won produces a specific urge, which is to immediately play another.
That's the worst possible moment to train or to play. Your judgement is off, your patience is gone, and you'll lose again.
The habit worth building is a hard stop after two losses in a row. Close the tab.
It sounds trivial. Players who do it gain rating from that rule alone, because most rating loss happens in the third and fourth game of a bad session.
The physical tell is worth learning. If you're moving faster than usual and your heart rate is up, you're tilted whether you feel angry or not.
Separate study from play
These are different activities and mixing them makes both worse.
Study is slow, deliberate and often frustrating. Play is fast, emotional and fun.
If you study when you meant to play, you'll rush the study. If you play when you meant to study, you'll learn nothing.
Put them in different slots, even different days. It removes a small decision that otherwise gets made badly every time.
A useful split is study on weekday mornings and play in the evening, or study on the days you're fresh and play on the days you're not.
How hard should the puzzles be?
Hard enough to fail some of them, easy enough not to dread it.
Failing about a third is the useful range. Failing most of them produces the feeling that ends training habits.
If you find yourself avoiding the session, the difficulty is a suspect before your motivation is.
Our guide to mate in one, two and three covers how to pick the right level honestly.
The rest day is not optional
Six days on and one day off works better than seven, and not only for motivation.
A day away lets the spacing effect do its job on everything you saw that week.
It also gives the habit somewhere to breathe. A streak that can never be broken becomes a source of stress, and stress is what ends it.
Pick the day in advance so it's a plan rather than a failure.
Signs you're already burnt out
Worth naming, because people miss them in themselves:
- You open the site and close it without doing anything.
- You play blitz instead of studying, every time.
- A loss ruins your mood for longer than the game took.
- You've stopped reviewing games because you don't want to see them.
Two or more of those means take a week off entirely. It works, and nothing else does.
A week off will not cost you rating
This is the fear that keeps people grinding, and it's unfounded.
Pattern recognition decays slowly. A week away costs you a little sharpness and returns most of it within two sessions.
What a week away does reliably is restore the appetite, and appetite is the resource that actually runs out.
Players come back from a break and play better within days. It surprises them every time.
Count sessions, not rating
Rating is a terrible progress measure over short periods because it moves for reasons that aren't about you.
So track something you control. Number of sessions this month is a good one.
It turns a bad week into a smaller number rather than into a failure, which matters more than it sounds. A missed week on a rating graph looks like decline. On a session count it looks like four instead of six.
I read the description of the forgetting curve while checking the spacing research, and it makes the same point from the other side. What decays is retention without review, not ability.
If the sessions are happening, the rating follows on its own timescale.
What to do with a plateau
A flat rating is not a reason to train harder. Usually it's a reason to train differently.
More of the thing that stopped working is the instinct, and it's wrong.
Changing what you study is almost always better than increasing how much, and it's far less likely to burn you out.
We went through the specific changes in breaking out of a rating plateau.
Make the session smaller than you want
If there's one thing to take away, it's this.
Decide on your session length, then halve it. Do the halved version every day for a month.
You'll finish most days wanting to carry on, which is the feeling that makes tomorrow's session happen.
That's the entire trick, and it works in every skill anyone has ever studied.
One other person makes it stick
The cheapest way to keep a habit going, and it costs nothing.
Tell one person what you're doing and send them your session count each week. That's the whole mechanism.
It works because a habit nobody can see is easy to drop quietly. A habit somebody asks about is not.
Better still if they're doing the same thing, because then you're comparing session counts rather than rating, and session count is the number you actually control.
What to do now
Pick a length you'd manage on your worst day. Pick a rest day. Write both down.
Set a rule for stopping after two losses in a row, and keep it even when it feels unnecessary.
Then do the small version for a month without judging it.
When you're ready to use what you've built, play a slow game on the play page rather than a fast one.