Quick answer
Quick answer: A plateau almost always means you're practising the thing you're already good at. Playing more games doesn't fix it, because playing is where you use skills rather than build them.
The four changes that move it are slower time controls, reviewing your own losses, endgame study, and cutting the opening reading.
I counted where my own study hours were going when I stalled, and three quarters of them were openings. That's the usual shape.
What a plateau actually is
Your rating stops moving when you've stopped meeting positions you handle badly.
At any given level you keep playing the same opponents and reaching the same shapes. You get better at those shapes, which is why you don't lose more. You don't get better at anything else, which is why you don't win more.
So the rating sits still.
It isn't a ceiling on your ability. It's a ceiling on what you're currently being exposed to.
Six months is normal, two years is not
Ratings move in steps, not in lines. A flat six months followed by a jump of a hundred points is the usual pattern rather than an exception.
So don't panic at a quiet spring. Look at the year.
If you've been within fifty points of the same figure for two years while playing regularly, that's a real plateau and something in your routine needs changing.
The rating system only reflects results, so a flat line means your results genuinely haven't changed.
Change one: play slower
The single most common cause is a diet of blitz.
Fast games train pattern recognition and nothing else. They don't train calculation, they don't train planning, and they punish the deliberate thinking you're supposed to be building.
One slow game a week teaches more than fifty blitz games.
If you only change one thing on this page, change this, and give it a month before judging.
Ten minutes with an increment is the minimum worth calling slow. Three minute games are entertainment rather than training, and there is nothing wrong with that as long as you know which one you are doing.
Change two: review your losses
Not with an engine first. With your own head.
Play through the game and mark the moves where you didn't know what to do. Those moments are the syllabus and nothing else is.
Then turn the engine on and see what it says about those exact moves.
Most players do this backwards, run the engine immediately, look at the biggest red bar, and learn nothing. We set out the better order in analysing your games.
Change three: study endgames
This is the least fashionable advice and the most reliable.
I compared where the hours go against where the losses come from. Endgames are the part of chess with correct answers, so studying them converts directly into points. And almost nobody at club level does it, which means the edge is unusually large.
Twenty minutes a week is enough. Rook endings first, king and pawn second.
We put the order in what to study first in the endgame.
Change four: stop reading openings
Opening study feels productive and is mostly wasted below expert level.
You learn twelve moves of a line, your opponent plays something else on move four, and you're on your own in a position you don't understand.
What you actually need is a handful of lines you understand the ideas behind, not a repertoire you've memorised.
Our openings trainer is built around that difference, and the ideas are what stick.
The puzzle trap
Puzzles are good and they're the easiest thing to do badly.
Solving a thousand easy puzzles builds speed at patterns you already know. That's the plateau in miniature.
The useful puzzle is one you can't solve in thirty seconds and have to work out properly.
So raise the difficulty until you're failing about a third of them. We went through the volume question in the fifty puzzles a day piece.
And stop when you get tired. Ten puzzles solved properly beat fifty guessed at, and guessing is exactly what a tired brain does.
Are you actually playing enough
The opposite problem is real too.
Some people study constantly and play twice a month, then wonder why nothing moves. Skills need using or they don't consolidate.
A reasonable split is two thirds playing and reviewing, one third studying.
If you're below about eight serious games a month, more study won't help and more games will.
Count the serious games rather than the total. Thirty blitz games in an evening is one session, not thirty data points.
Check what your losses have in common
Go through your last twenty losses and sort them into three piles.
Blundered a piece. Got outplayed positionally. Lost a won or drawn ending.
Whichever pile is biggest is your whole training plan for the next two months. Everything else can wait.
Most people never do this and study whatever's in front of them, which is why the plateau holds.
If two piles are the same size, take the one that appears in your most recent games. That is the one you are still doing.
If the blunder pile is biggest
Then your problem is a process problem rather than a knowledge problem.
Run the same two checks before every move. What did their last move attack, and what does my move leave undefended.
It's boring, it works, and it's worth more rating points below fifteen hundred than any other change.
And it's the reason slower time controls help so much, because the check needs time to happen.
If the positional pile is biggest
Then you need plans rather than moves.
The fix is studying pawn structures, because the structure tells you what the plan is. Isolated pawn, hanging pawns, a closed centre, opposite side castling. Each has a standard plan and knowing three of them changes how you play.
This is slower going than tactics and it's the difference between a thirteen hundred and a sixteen hundred.
Pick one structure and play it from both sides for a month.
If the ending pile is biggest
Then you're losing points you'd already won, which is the most annoying and the most fixable.
Start with rook and pawn against rook, because rook endings are the most common ending in real chess by a long way.
Then king and pawn oppositions, then the basic mates.
That's maybe six hours of study total, and it stops a category of loss permanently rather than gradually.
Play stronger opponents on purpose
A plateau is partly an exposure problem, and the direct fix is playing people who beat you.
You'll score less for a month or two. You'll also meet positions your usual opponents never create, which is exactly what stopped happening.
One event or one session a week above your level is enough.
Expect the rating to dip before it moves, and don't read the dip as failure.
The bots are useful here because you can set the strength exactly and repeat it, which a human opponent cannot do for you.
How long to give it
Three months before you judge any of this.
Ratings are noisy and a month of results tells you almost nothing about whether a change worked. Two hundred games is a fair sample. Twenty is not.
So pick one change, commit to a season of it, and stop checking the number weekly.
Checking the rating after every game is itself a reason people stall, because it turns practice into performance.
When it is genuinely time
Sometimes the honest answer is that you're playing enough and studying enough and it's going to take longer than you hoped.
Chess improvement is slow above a certain point and the curve flattens for everyone eventually.
That's not a reason to stop. It's a reason to pick a longer horizon and enjoy the games.
A plan with a year in it beats a plan with a month in it, and it makes the quiet stretches much easier to sit through.