Quick answer
Quick answer: Mate in one trains recognition. Mate in two trains the idea of a forcing move. Mate in three trains search, which is a different skill and much slower to build. Drill whichever one you currently fail, not the hardest one you can solve.
Most players jump to mate in three too early and learn nothing except frustration.
I counted the named mating patterns in the standard reference. There are forty, and about six of them account for most club games.
Why mate in one still matters
Strong players dismiss these. They shouldn't.
A mate in one puzzle is a pure recognition test. Either the pattern is in your head or it isn't, and there's nowhere to hide.
What you're building is the ability to see a finished mating picture at a glance, without calculating.
That's the foundation for everything else, because a mate in three is just a mate in one with two moves of setup in front of it.
What mate in two actually teaches
The jump from one to two is bigger than it looks.
Now you need a first move that forces a reply, and forcing is the whole idea. Checks, captures and threats.
Most mate in two puzzles have a quiet first move that looks wrong, which is exactly why they're instructive.
If you can solve mate in one instantly but stall on mate in two, your gap is forcing moves rather than patterns.
Why mate in three is a different sport
At three moves you cannot check every option, so you need a way to choose which to look at first.
The arithmetic is unforgiving. With around thirty legal moves a side, three moves deep is a tree with tens of thousands of branches, and nobody looks at those.
So what you're really training is pruning. You learn which shapes are worth investigating and you ignore the rest without noticing you did.
That's built from experience rather than from rules. It's why mate in three feels impossible until it suddenly doesn't, usually after a lot of mate in twos.
Trying to skip to it is like doing long division before you know your times tables.
The patterns worth knowing first
I read through the whole list, and most of it is beautiful and rare. These five are the ones that turn up in real games:
- Back rank mate, where the king is trapped by its own pawns.
- Smothered mate, delivered by a knight to a king surrounded by its own pieces.
- Ladder mate, two rooks walking the king to the edge.
- Boden's mate, two bishops on crossing diagonals.
- Arabian mate, a rook and knight working together in the corner.
Learn those five and most of your missed mates disappear.
The rest are worth browsing once for enjoyment. Names like Opera mate and Blind swine mate come with stories attached, and a pattern with a story sticks better than a diagram on its own.
Back rank first, always
If you only learn one, learn this one.
A back rank mate happens because the king's own pawns block its escape, and it comes up in club games constantly, on both sides of the board.
Knowing the pattern does two things. You spot it when it's available, and you spot it when it's about to happen to you.
The second is worth more. A move to give your king air is the cheapest insurance in chess, and it costs you one tempo in a quiet moment.
The habit to build is a glance at your own back rank every time the queens are still on and your rooks leave home. That glance takes a second and it saves whole games.
How to actually practise
The method that works is boring and specific.
Solve without moving the pieces. If you have to try a move to see what happens, you're testing rather than calculating, and testing doesn't build the skill.
Set the position, find the answer in your head, then check.
That single rule turns puzzle practice from entertainment into training.
Time limits change what you learn
Timed puzzle modes are fun and they train a specific thing, which is speed of recognition.
They do not train calculation, because there's no time to calculate.
So a timed run is good for mate in one and increasingly useless as the puzzles get deeper.
We compared the two popular timed formats in puzzle storm versus puzzle rush.
The mistake that wastes the most time
Solving puzzles you already know how to solve.
It feels productive, the accuracy stays high, and nothing changes. A puzzle set where you score ninety percent is entertainment.
You want a difficulty where you fail about a third of them. That's uncomfortable and it's where the learning is.
The reason people avoid it is that puzzle sites show you a rating, and failing makes the rating go down. So the incentive built into the tool points away from the thing that helps you.
If your puzzle rating has been flat for a month, the difficulty is almost certainly too low.
What to do when you can't see it
Work through the forcing moves in order rather than staring:
- Every check you can give, including ones that look silly.
- Every capture.
- Every move that threatens mate next.
In a mate in two or three, the answer is always in that list. Staring at the position is not a method.
Why sacrifices look wrong
Most composed mating puzzles begin with a move that loses material.
That's deliberate. The whole point is to break the habit of counting pieces before looking at the king.
In a mating net material means nothing. A queen that delivers mate is worth exactly as much as a pawn that delivers mate.
Once that clicks, a whole category of puzzles stops being hard.
Composed puzzles versus real positions
There's a difference and it's worth knowing.
Composed puzzles have exactly one solution and it's usually elegant. Positions from real games often have several winning moves and one that's fastest.
That changes how you should think. In a composed puzzle, a move that merely wins a rook is wrong. In your own game it might be the sensible choice.
Composed puzzles train precision. Game positions train judgement.
Do both. The game positions are the ones that transfer, and the composed ones are the ones that teach patterns quickly.
Forcing mates you should just memorise
Some endings are not puzzles at all, they're technique, and you either know them or you lose half points.
King and queen against king. King and rook against king. Two bishops against king.
Those are procedures rather than puzzles, and an hour spent on them pays for years.
We worked through the hardest common one in forcing mate with two bishops.
How many to do a day
Fifteen minutes, every day, beats two hours on a Sunday.
Pattern recognition is built by repetition spaced over time, and the gap between sessions is part of the mechanism.
Ten puzzles daily is plenty if you solve them properly rather than guessing.
Guessing quickly through fifty teaches you almost nothing, and it's the default behaviour on every puzzle site.
Say the move out loud before you play it
It sounds silly. It works, and there's a reason.
Saying the move forces you to finish the thought, including the part where you check what the defender can do about it. Clicking lets you guess.
So the habit is short. Name the move, name the reply you expect, then play it.
Most failed puzzles aren't a failure to see the idea. They're a failure to check one legal defence, and speaking the line catches that defence far more often than staring does.
What to do now
Find your level honestly. Do ten mate in twos and count how many you get without moving a piece.
If you get eight or more, move to mate in three. If you get four, stay where you are and learn the five patterns above.
The full list of named patterns sits in the checkmate pattern reference if you want to browse them.
Then play a game and look for one of them deliberately. You can start on the play page.