Article

Published 6 August 2026 by Reini Wust, who builds Chessou.

What your chess game review can't tell you

You read the review. You saw the move you missed. Two games later you make the same mistake again.

That is not a discipline problem, and it is not a memory problem. A review tells you what happened in one game. Whether it changes the next one depends entirely on what you do after you close it. Learning research has measured that gap for decades, and the findings are unusually clear.

Why does the same mistake keep coming back?

Because knowing better was rarely the problem.

In 2008, the cognitive scientists Bilalić, McLeod and Gobet gave strong players positions that held a familiar winning idea and a better one hidden behind it. The familiar idea was enough to hide the better one. Players who handled comparable positions without trouble suddenly performed like players far below their own strength.

Five years later, Sheridan and Reingold tracked where those players were actually looking. They kept their eyes on the familiar move while they believed they were still searching for something better.

Time pressure does the rest. Chabris and Hearst went through hundreds of games by 23 grandmasters: faster play produced clearly more and bigger errors than slow play. Knowing a position inside out does not protect you when the clock is short.

Your review can explain the position perfectly. Next game, the familiar move is still the one you see first.

In Chessou a pattern that keeps coming back is not left to the single game. Chessou knows every game you have played, points out that this one has been showing up for a while, and offers to turn it into a drill on the spot.

Don't review forty moves. Review three key moments.

Everyone tells you to analyse your own games. Almost nobody tells you how, and that is where most people stop. An engine puts a number next to every single move, and that still does not tell you where to start looking.

A game is usually decided in two or three positions. Finding those is the whole job. Everything else is moves where nothing was at stake.

Do it that way and a review takes a few minutes instead of being put off again, and those minutes go into the moments that decided the game.

This is not a Chessou idea. GM Noël Studer, who works with improving players, teaches the same thing: focus on the critical moments instead of walking through everything. He also names the trap on the other side of it, that you will only ever find what you are looking for.

In Chessou you do not have to find them yourself. Every game is analysed on arrival, and you are offered a guided run through its three most important key moments, one at a time.

Knowing how to review a game is mostly knowing which three key moments to look at.

What is a game review, really?

It is feedback. And feedback has been measured for decades.

Van der Kleij, Feskens and Eggen pooled 40 studies of computer generated feedback. Simply telling someone that they were wrong barely moved their later performance at all. Feedback that showed how to do it differently moved it roughly ten times as much.

Kluger and DeNisi went further and pooled 607 measured effects. After feedback, performance got worse in more than a third of cases. Being told about your mistakes is not automatically good for you.

An engine label sits at the weak end of that scale. It marks the move and gives you a number. It is accurate and it is instant, and on its own it is close to the least effective form of feedback there is. That is true of every game review, the one in Chessou included.

In Chessou every move gets a comment, not just an evaluation. Tactically and strategically, so you find out what the position was asking for and not only that something went wrong.

What turns a review into a change

Doing it yourself, afterwards.

Stephen Wong ran the comparison directly in 2023. People who made the error themselves and then got the correction remembered more, and carried more of it over to new problems, than people who were handed someone else's error and asked to correct it. Spotting a mistake is not the same as having made it.

Potts and Shanks found the same thing with people who had to answer before they were told. They spend less time with the right answer than people who just read it, and they still remember it better afterwards.

Reading about a mistake is the weaker version of both. Playing the position again is the stronger one.

Why it should be a position from your own game

Because of which errors get corrected best.

The mistakes people fix most reliably are the ones they were sure about. Being wrong when you were confident is surprising, and the surprise is what makes the correction hold. Janet Metcalfe collected two decades of work on this, and Carpenter and colleagues found it again with real course material rather than quiz questions. Answers you were unsure of are the ones that quietly come back.

A position from a stranger's game cannot do that for you. You never had any conviction in it. The move you played yourself two days ago is the one you were certain about at the time.

In Chessou there is no general puzzle pool to fall back on. Every position you train is one you reached in a game you played.

Worth being straight about: this is the mechanism, not a scoreboard. No one has run the direct comparison between training on your own games and training on a general puzzle set.

Why seeing it once is never enough

You understood the mistake. That does not mean it is gone.

What keeps a correction from wearing off is meeting it again after a few days, and once more a week later.

Karpicke and Roediger showed how big that difference is. Two groups learned the same material. One group went through it all in a single sitting. The other came back to it several times, with days in between. A week later, the first group remembered 36 percent of it. The second remembered 80 percent.

For your chess that means the mistake you made on Tuesday has to be put in front of you again on Friday, and again the week after. Reading about it once, on the evening you made it, is the version that fades.

What to do after your next game

Three steps, and none of them need Chessou. They follow from everything above.

1. Go through the game once before you turn on the engine. Mark the two or three points where you were unsure, or where you remember the game turning. You are looking for the moments, not for the moves yet.

   In Chessou every move is analysed the moment the game arrives, and the review opens on up to three key moments, picked out of the whole game and lined up in order.

2. At each one, find your move again before you read the answer. Say it out loud or write it down, then compare. Reading the answer straight away is the version that does not stick.

   In Chessou those positions come back to you as puzzles, so you play them instead of reading them.

3. Put the position in front of yourself again a few days later. Once is where most corrections quietly come undone.

   In Chessou that is scheduled for you, and each position comes back right before you would forget it.

The first step is the one people skip, and it is the one that decides whether the other two are worth anything.

What happens in Chessou after the review

Chessou reads your games from Lichess and Chess.com on its own, and you can paste in a PGN for anything you played elsewhere, over the board included. Every move is analysed with native Stockfish. The review is where that becomes readable. Everything below is what happens next, without you setting any of it up.

The review is the start of the training, not the end of the game.

Try it on your own games

Enter your Lichess or Chess.com username and your games are imported and analysed before you pay anything. Every account starts with one week free, everything included. Plans are on the pricing page. There is also a piece on the chess blunder you keep repeating, and a side by side with Chess.com if that is where you play.