You just finished a game. What do you do next?
If you're like I was for about four years, you hit "rematch" immediately, or you scroll the engine's eval graph for thirty seconds, squint at the move where it tanked, and close the tab like you've learned something.
That's not analysis. I don't know what it is. Post-mortem theater, maybe. It looks active. You walk away with nothing your next game can use.
Real analysis, the kind a decent coach charges eighty bucks an hour for, is a structured process with phases. Each phase answers a different question. Skip a phase and the whole thing collapses into "yep, that was a blunder, noted."
I spent a year getting this wrong. Then I spent another year figuring out how coaches actually do it. What follows is the five-phase method I ended up with. Use it tonight on your next loss. It takes twenty to thirty minutes per game, which is honestly less time than you've already spent staring at engine evals that didn't stick.
Why most players analyze wrong
Most of us (and I include past me here, heavily) make the same handful of mistakes.
Going to the engine first. The engine tells you the answer. Once you know the answer, you can't engage with the question. You were about to work out what you missed, and now you're just nodding at Stockfish. The learning happens in the gap between your guess and the engine's. If you never make the guess, there's no gap.
Only analyzing losses. Your wins contain lessons too. Sometimes you won despite playing poorly, which sets you up to lose the exact same way in the rematch. A good win is one where you played well, not one where your opponent played worse.
Stopping at "I hung a piece." Why did you hang it? Time pressure? Calculation failure? You just didn't see the threat? Three different failure modes, three different fixes. "Blundered" isn't a diagnosis.
And not writing anything down. If you don't externalize the lesson, it's gone by Thursday. The notebook is the thing that makes analysis stick, not the analysis itself.
Last one: shallow analysis of many games rather than deep analysis of a few. Twenty games skimmed teaches you less than two games dissected. Took me three years to accept this.
Phase 1: The post-game reaction (skip this, but know why)
Right after a game you're emotional. You just won or lost. Adrenaline is still settling, and your read on the game is wildly unreliable.
If you lost, you'll be too harsh on yourself and miss two things you did well.
If you won, you'll skip past the three moves where you almost threw it.
Don't analyze now. Wait at least an hour. Ideally sleep on it.
The one exception is tournament play, where you don't have twelve hours before the next round. In that case, just jot down the critical moment or the thing bugging you, close the game, and save the real analysis for after the event.
Phase 2: First pass — play through without judgment
Open the game. Engine off. Click through every move in order.
Don't evaluate. Don't stop to think about alternatives yet. Just re-experience the game. At each move, ask yourself, what was I thinking? What was I trying to do?
Pay attention to your clock. Where did your time go? Which moves did you think on for a long time, and which did you click through in three seconds? Both matter. Long thinks mark where you struggled. Fast moves mark where your evaluation might have been too shallow for the position.
As you go, mark (mentally, or with an annotation if your site supports it) three to five moments where you paused hardest or where you felt the game swing. These are your critical moments.
About ten minutes. You're not learning anything yet. You're loading the game back into working memory so the next phase has something to bite into.
Phase 3: Critical moments — your thinking before the engine
Now, for each critical moment, write down, actually write down, four things.
- What did I see in this position?
- What candidate moves did I consider?
- Why did I pick the move I picked?
- What do I think I missed?
Do this for every critical moment before touching the engine. This is the phase most players skip, and it's the one that matters.
Why? Because improvement happens in the gap between your guess and the truth.
If you write "I picked Nxf7 because I thought it won a pawn with tempo, I didn't look carefully at Bb4+," and then the engine confirms Bb4+ was winning, you now know something about your own calculation habits. You stop calculating once you see an attractive idea. That's diagnostic. That's a training target.
If you skip to the engine first, you see "Bb4+ = +2.8" and shrug. You learn a move, not a pattern about yourself.
Phase 4: Engine pass — verify and tag patterns
Engine on. Full analysis. Walk through the critical moments again with the evaluation visible.
For each blunder or missed opportunity, tag it with a pattern name. Keep the tags short and concrete:
- fork I missed
- back-rank weakness I ignored
- overloaded defender
- initiative surrendered
- bad trade in endgame
- didn't calculate opponent's threat
- played too fast in equal position
- thought I was lost, gave up
These tags matter because patterns repeat. If you analyze five games and three of them show "missed knight fork," you've got a clear training target. Drill those positions specifically. If six games in a row tag "didn't calculate opponent's threat," that's your thinking process, not your tactical vision. Different fix.
The tags are the difference between "I blundered a piece" (useless) and "I blunder pieces specifically when my opponent has a forcing reply I didn't visualize" (actually trainable).
Phase 5: Extract the lesson
Every game analysis ends with exactly one sentence. Written down somewhere you'll read it later.
Examples:
- I keep missing tactical motifs when my clock is under 3 minutes.
- I trade down too aggressively when I'm up material.
- I play too fast in equal endgames.
- When I'm Black against e4 I get a worse position straight out of the opening.
One sentence. If you can't reduce the game to one, go back to Phase 4 and look at your tags again.
Keep the lessons in one running document. Apple Notes, Notion, a physical notebook, whatever. Review it weekly. The sentences pile up into a portrait of your weaknesses, and that portrait is what you actually train against.
Common mistakes when applying this
A few ways people (me) blow this up even after learning the method.
Spending two hours on one game. The whole framework should be twenty to thirty minutes. If you're at the hour-plus mark, you're overthinking individual moves. Stop.
Analyzing every game. Pick two or three per week to go deep on. The others, just tag critical moments on a quick pass.
Only analyzing losses. You already know. Your wins contain lessons.
Getting hung up on the engine's top line. The engine is a tool, not the answer. If Stockfish's top move is incomprehensible to you, your training target is the gap between your best move and the engine's, not memorization of engine lines you don't understand.
Tools that help
You can do all of this with a notebook and any game analysis site. Chess.com's game review and Lichess's computer analysis both work for the engine pass.
For the notes layer (the critical-moment writeups, the lesson list), Apple Notes, Notion, a physical notebook, whatever you'll actually open. Mine was a Moleskine for about ten months because I'm a cliché.
There are a few chess apps that automate parts of this. Chessy (full disclosure, I use it) does Phase 4's pattern tagging automatically, generates puzzles from the critical moments it finds, and keeps your lesson list in one place. Optional. I did it on paper for most of a year before switching, and the paper method works.
How often
A reasonable cadence:
- Two or three deep analyses per week (all five phases)
- Quick tag-only passes on everything else (just mark critical moments and patterns, don't go deep)
- Weekly review of your running lesson list, looking for repeat patterns
Maybe an hour and a half per week of structured review. Less than the time you spend playing.
One more thing
The difference between a 1200 and an 1800 isn't talent. It isn't hours logged. I've seen 1200s who play more than 1800s and it hasn't closed a point. The difference is whether they have a method for learning from their own games.
An 1800 has a method. A 1200 plays more and hopes.
You can do this tonight. Open your last loss. Walk through it with the engine off. At the three hardest moments, write down what you were thinking.
That's analysis. Everything else is looking at moves.