Most adult improvers do their “game analysis” like this: lose a game, paste it into the engine, watch the evaluation bar swing, nod at a red blunder marker, close the tab. Twenty seconds. Nothing changes. Three weeks later you hang the same piece in the same kind of position and wonder why your rating won’t move.
The problem isn’t effort — it’s method. An engine tells you what the best move was. It almost never tells you why you didn’t find it, and that “why” is the only thing that transfers to your next game. After reviewing hundreds of amateur games, I’ve found the improvers who climb fastest all do roughly the same thing: they analyze their own thinking first, and let the engine confirm or correct it second. This is the repeatable, ~15-minute method I give them.
Why Engine-First Analysis Quietly Fails
When you open the engine before forming your own opinion, you outsource the hardest and most valuable part of analysis: the act of judging a position yourself. You see the +2.3 and the recommended knight move, and your brain files it as “ah, of course” — a feeling of understanding that evaporates the moment the engine is gone. Cognitive scientists call this the fluency illusion: recognizing a correct answer feels identical to being able to produce it, but they are completely different skills.
There’s a second trap. The engine flags the objectively biggest swings, but a –4.0 blunder you’d never repeat is far less useful to study than the quiet –0.6 inaccuracy you make in every single game. Improvement comes from fixing recurring, systematic leaks — not from gawking at your worst one-off disaster. If you want a deeper read on what those evaluation numbers actually mean, our centipawn guide for adult improvers breaks it down, but the headline is this: the number is a starting question, not the answer.
The 5-Step Self-Analysis Method
Do steps 1 through 3 with the engine completely off. Only turn it on at step 4. This sequence is the entire trick — everything good follows from analyzing before you verify.
Step 1 — Replay and narrate the game from memory, engine off
Open the game and click through your moves while talking (out loud or in writing) about what you were thinking at the time. “Here I wanted to trade the bad bishop.” “Here I had no idea what to do so I just developed.” The goal is to surface your actual reasoning, including the gaps. The moments where your honest narration is “I don’t know why I played this” are gold — those are your real weaknesses, and you just found them without any software.
Step 2 — Mark your three critical moments
Don’t analyze all 40 moves; you’ll burn out and learn nothing. Find the two or three turning points — where the game’s character changed: a missed tactic, the moment you lost the thread, a pawn break you allowed, the transition into an endgame you didn’t understand. Most decisive amateur games hinge on two or three real decisions. Mark them and ignore the rest.
Step 3 — At each critical moment, write down what you’d play now and why
This is the rep that actually builds skill. Sit on each marked position and commit to a move and a one-sentence reason before the engine speaks. Force yourself to produce a plan: what are you trying to do here? If you struggle to generate any plan at all in quiet positions, that’s a specific, trainable gap — our walkthrough on how to find a plan in chess gives you a checklist for exactly these moments.
Step 4 — Now turn on the engine, and compare reasoning, not just moves
Switch the engine on and check your step-3 answers. But don’t stop at “right” or “wrong.” For every gap between your move and the engine’s, ask the diagnostic question: what category of mistake was this? Was it a calculation slip (you saw the idea but miscounted), an evaluation error (you misjudged who stood better), a missed candidate move (the right idea never entered your head), or a plan failure (you had no direction)? The category matters far more than the specific move, because the category is what repeats.
Step 5 — Log one sentence in a recurring-mistakes journal
Keep a single running document. After each analysis, add one line: the mistake category plus the trigger. “Missed opponent’s knight fork because I didn’t check his replies after my last quiet move.” Over ten games, patterns scream at you — and a pattern you can name is a pattern you can train. This log, not any single brilliant find, is where the rating points come from.
What Your Logged Patterns Are Actually Telling You
After eight to ten logged games, your journal stops being a list of mistakes and becomes a profile. Maybe 70% of your entries are “didn’t check opponent’s forcing replies” — you have a calculation discipline problem, not an opening problem, and no amount of new opening theory will fix it. Maybe they cluster in positions where you’re slightly worse and start drifting — an emotional and resilience leak more than a technical one. If your losses pile up right after an earlier loss, the issue may not be chess at all; it’s tilt, and our post-loss recovery system addresses that directly.
This is the same logic behind every plateau being beatable: the leaks holding back a 1600 are specific and few, not vague and infinite. We mapped the common ones in the three quiet leaks holding back 1600–1800 players. Self-analysis is simply how you discover which of those leaks is yours, rather than guessing.
How Often, and How Long
You do not need to analyze every game. One deep 15-minute review beats five frantic engine-skims. A realistic, sustainable cadence for a working adult: analyze every loss properly, plus the occasional win that felt confusing. Two or three real reviews a week is plenty. The non-negotiable part is step 5 — if you analyze but never log, you’ll rediscover the same lesson forever. The log is the compounding asset.
Where a Personalized Plan Fits In
Self-analysis works, but it has one honest limitation: you’re the same person who made the mistakes, so your blind spots are, by definition, hard to see in your own games. That’s the gap MyChessPlan was built to close. Upload your games and our analysis reads across all of them at once, surfaces the patterns you can’t see from the inside, and maps them to your player archetype.
You can start completely free: our free archetype report tells you what kind of player your games say you are and where your most expensive recurring leak is. If you want the full fix, the $14.99 personalized improvement plan turns that diagnosis into a concrete training sequence — the specific calculation drills, endgames, and opening fixes aimed at your logged patterns, not generic advice. Think of it as the recurring-mistakes journal from step 5, built for you automatically and turned into a study plan.
Get your free archetype report →
The Bottom Line
Stop letting the engine think for you. Replay from memory, find your two or three critical moments, commit to a move and a reason before you verify, classify the mistake by category, and log one sentence every time. Do that for ten games and you won’t need anyone to tell you what to work on — your own journal will. The engine is a fact-checker for your judgment, not a replacement for it. Build the judgment first.