There is a version of game analysis that goes like this: you open your last game, watch the engine line climb from +2 to +8, sigh, and close the tab. Nothing was learned, and you feel slightly worse about yourself. The goal here is to replace that with a routine that takes fifteen minutes and actually sticks.

Start before you open the engine

This is the step almost everyone skips, and it is the most valuable one. Before you let Stockfish say anything, replay the game from memory. For the moves that surprised you, stop and ask what you thought was happening at the time. Did you believe you were winning when you were losing? Did you feel cramped the whole game but not know why?

Write those moments down, even as a couple of bullet points. The point of analysis is not to produce a perfect scorecard; it is to notice the gap between what you intended and what actually happened on the board.

Find the turning point, not the blunder

Every game has one move where the evaluation swings hard in one direction. Often it is not the flashy queen sacrifice — it is the quiet pawn move on move 14 that you made without thinking. Use the move list in the Replay tab and watch the evaluation bar change. When it drops sharply after one of your moves, that is your real lesson.

Ask what the position wanted before you asked the engine. A common pattern: you traded pieces when you were better, and the game slowly dried up into a draw. The engine will show you the evaluation drift, but you have to recognize the moment yourself.

Read the suggested move, then understand it

When the engine shows a better move than yours, the temptation is to memorize it and move on. Instead, put the move on the board and look at the position after it. What did the engine gain — material, space, king safety, a structural weakness in your opponent's position? The best move usually follows a one-sentence idea like "this pawn becomes weak" or "the king is stuck in the center."

If you can say that sentence, the lesson transfers to future games. If you cannot, it probably will not.

Avoid the blunder-check spiral

Once you find one bad move, the engine will cheerfully find twenty more. That does not mean you played twenty bad moves. Classical engines evaluate every single move at the same depth, so a slightly-imprecise move 3 gets labeled the same way as a losing blunder on move 30. In this tool the classification thresholds are fixed, so a "mistake" in a winning position can look worse than it felt at the time.

Take the two or three moves with the biggest practical cost, understand them, and stop there. Analysis is a scalpel, not a flamethrower.

A simple routine

Something like this works well for most club players:

1. Replay the game from memory and mark the move that surprised you.
2. Find the one evaluation swing and understand what changed on the board.
3. Understand the engine's suggestion for that move.
4. Note one thing to try next game — not three things, one thing.

That is it. Fifteen minutes, one lesson, no shame spiral. Do that for twenty games and your chess will change more than from reading a hundred opening videos.