Almost nobody reviews their own games, and those who do review them badly: they open the replay, watch it end to end, get angry a second time about the same deaths, and close the client convinced they've studied. Review that works is short, targeted, and produces one thing: a correction to carry into the next game. This article is the method, with the minutes counted.
One clarification. The game numbers quoted here (objective timers, minion values, wave cadence) are current as of patch 26.15. The minutes of the method aren't game data: they're operational guidance, calibrated to keep the review inside a real half hour.
Why naive review doesn't work
Three ways to get it wrong.
Watching everything. For the first few minutes you're focused; the rest of the time you're watching TV. A full rewatch isn't analysis, it's content consumption, and it leaves nothing.
Hunting for blame. When you reopen a loss, your brain already knows what it wants to find: proof it wasn't your fault. The jungler never came, the top laner lost his lane in five minutes. It may be true, and it's useless: it isn't a variable you control. The only question that produces improvement is: with that team and that information, what could I have done differently?
Stopping on the spectacular deaths. The dive where you die three against one is memorable, but it's rare and doesn't come from a pattern. The death that costs you the season is the boring one: minute twenty-two, side lane, nothing happening, you crossed the river for no reason and died. You don't even remember it, and you repeat it every game.
Which game to pick, and which to avoid
Selection matters more than the method. Two immediate discards.
Not your worst loss. The signal is buried under the noise: everything went wrong, so anything you look at looks like a cause. You're still angry too, and reviewing hot is a second lap of frustration. Without a way to switch off, read How to Stop Tilting in League of Legends: Mental Game Guide first.
Not the clean win. The feedback is positive throughout: whatever you did "worked", including the mistakes with no consequences.
Two candidates are good: an ordinary loss that stayed close and where something slipped, and a win in which you felt passive, with acceptable stats and no real influence. In the first the signal is isolable; the second teaches more, because the positive result hides the problem.
On timing: within a day or two, never straight afterwards. You need memory of your intention and enough distance to contradict yourself. There's a technical constraint too: the replay has to be downloaded from your match history, and stays playable only until the next patch, roughly two weeks. If the game sits right before a patch, download it now.
The budget: thirty minutes, six blocks
An operational split; the total has to add up.
- 5 minutes: picking the game, downloading the replay, one fast pass noting the minute of each death.
- 6 minutes: the death timeline, opening only those moments.
- 6 minutes: the early game, in four frames.
- 5 minutes: the hunt for passive minutes.
- 5 minutes: one map decision, reconstructed in full.
- 3 minutes: choosing the correction and writing the note.
Thirty exactly. Keep a real timer: when you overrun, it's because you stopped analyzing and started watching.
The death timeline: three blunt questions
Build the list of minutes you died in yourself: open the replay, raise the playback speed to the maximum, note every death in one pass. It's the only manual step, and at 8x a thirty-minute game runs through in just under four: that's why the first block is worth five.
Step zero, before you reopen the first death: make sure the replay is showing the map with your team's vision only. On full vision the minimap reveals enemies you couldn't see live, "what did I know" gets a false answer, and you inflict hindsight bias on yourself.
Then don't rewatch the deaths in the flow of the game: jump to those minutes and start twenty seconds before. The point isn't the death, it's the decision that took you there. For each, three questions and short answers.
- Where was I? Not "in mid": relative to the wave and the nearest turret, was I in front or behind, and did I have a way out?
- What did I know? Which enemies were visible on the minimap in the previous ten seconds. Did I have a live ward on that side? Had I actually seen my opponent's recall, or assumed it?
- What was I gaining? CS, plates (120 gold split between everyone nearby; since patch 26.01 they no longer expire and count for as long as the turret stands), vision, the setup for an objective, or nothing.
If the answer to the third is "nothing", that death counts double: it's free, and free deaths are the easiest to remove, because they don't ask you to play better, only not to be there.
Then classify. Execution death: the decision was reasonable but you botched it, you missed the skillshot, burned Flash too late. Decision death: wrong place, wrong time, without the information you needed. The second is worth far more, because it's corrected with a verbal rule, applicable immediately and transferable to any champion.
The elo bracket changes the reading. The scale is always the same, three steps: Iron and Bronze; Silver and Gold; Platinum and up, meaning Platinum, Emerald, Diamond, and everything above. In Iron and Bronze most deaths look like execution and are decision deaths in disguise: you lost the trade not because you pressed the wrong key, but because you started it when your opponent had everything up and you didn't. In Silver and Gold they're nearly all wave-related: you die from staying in lane with the wave pushing toward you. Platinum and up they move into the mid game and are about positioning around objectives.
The early game in six minutes, not fifteen
The temptation is to rewatch the whole laning phase. Four frames are enough.
Frame 1, the first cannon. It leaves the base with the third wave, at 1:30, and reaches lane shortly after. It's worth 50 gold against the 20 of a melee and the 14 of a caster, and grows by 1 with every minion upgrade, so late in the game it's worth almost 70: it's the minion that costs the most to miss. Check whether you took it, and what state the wave was in.
Frame 2, the first recall. Planned or forced? A planned one starts with the wave pushing toward your opponent, sensible gold to spend, and a clear idea of the minute you get back. A forced one you take because you're low, and its cost appears in no end-of-game statistic.
Frame 3, the return after the recall. How many minions did you lose between the base and your next last-hit? XP is picked up within 1500 units of where the minion dies, with no last hit required: back inside that radius, you only lose gold. Further out, under your own turret while the minions fight mid-lane, you lose experience too, and that loss weighs more, because it moves your level 6. Manipulating the wave before you go back is covered in Wave Management in League of Legends: the Skill That Separates Iron from Diamond, last-hitting in CS Like a Pro: From 5 to 8 Minions per Minute.
Frame 4, the first interaction with the enemy jungle. The first gank you took, or dodged, tells you whether you were playing from a defensible spot. Whether you died doesn't matter: what matters is whether you had reason to expect it thirty seconds earlier.
The minutes when nothing happens
This is the most underrated block, and why plenty of players with decent stats stay stuck for whole seasons. A passive minute, operationally: an interval in which you don't take CS, don't push a wave, don't move toward an objective or a position to contest it, don't place or clear vision, and don't fight. You're just existing on the map.
The cost enters no statistic: your KDA stays clean, your CS mediocre but not catastrophic. The game hasn't stopped to wait for you, though. Waves come out every 30 seconds, at a cadence that speeds up to 25 seconds from minute 14 and to 20 from minute 30. Every passive interval is gold leaving the game, or going to your opponent.
How to measure them: run the replay at high speed and note the start and end of every interval in which you walk with no destination or stand still. Then compare. In the laning phase a wave is three melee minions at 20 gold and three casters at 14, so it's worth 102 gold, or roughly 152 with the cannon. That holds until minute 14: from there cannon waves lose a minion, and from minute 30 the composition changes again. At one wave every 30 seconds, a passive minute costs you on the order of two waves, plus experience. The number is yours, and more convincing than any average.
The recurring causes are few: waiting for a teammate who never arrives; watching a teamfight you won't join in time; standing in mid to hold prio without actually pushing.
One map decision, reconstructed in full
Five minutes, one decision. Not three, one. Pick a rotation or a contested objective: the first drake (bottom pit, at 5:00, then five minutes after each kill), the Void Grubs (at 8:00, once only, gone at 14:45, three stacks at most against structures), the herald (top pit, at 15:00, gone at 19:45), or the baron (at 20:00).
Then reconstruct the forty-five seconds before it, not the objective. Three things, in order.
- The state of the three lanes. Who was pushing into whom. An objective is taken when the waves are working for you, not when the team decides.
- The vision available. Not "we had vision", but which spots were covered, for how long, and who had placed a control ward.
- Where you were forty-five seconds earlier. Most objectives are lost before they start, because two players arrive ten seconds late.
The bracket decides whether the time is worth spending at all. In Iron and Bronze these five minutes pay off more on a second pass over the free deaths and your CS at minute ten: the decision there is almost always binary. In Silver and Gold the block starts to pay, on the state of the lanes more than on the timing. Platinum and up it's the block that makes the difference, and it's nearly always a question of setup: you reach the right position fifteen seconds late because of an unplanned recall.
The one-correction rule
By the end of the twenty-seven minutes you'll have seen at least five things to fix. Take one, and choose inside the final three minutes: pick the one that touches the most of what you saw. The reason isn't motivational, it's attention. Spread across three corrections, you revert to your default behavior on all three the moment the game heats up.
The correction has to be a verifiable sentence: "I play the lane better" isn't one. "I don't cross the river without a ward placed in the last thirty seconds" is, and so is "my first recall only starts when the wave is pushing toward them". In the games that follow, count how often you respected it and how often you broke it: the honest count is half the work. When you execute it without thinking for three games running, it has landed.
The correction depends on where you are, too. In Iron and Bronze a fundamental, measurable correction almost always wins: free deaths, CS at minute ten, control wards bought. In Silver and Gold wave and recall corrections pay. Platinum and up you need timing and objective-positioning corrections, because the fundamentals are no longer the limiting factor.
What to write, and where
Inside those same three minutes, five lines, always the same format.
- Line 1: date, champion, matchup, result.
- Line 2: the most representative death, with minute and classification.
- Line 3: the longest passive interval, and what you were waiting for.
- Line 4: the single correction, written as a verifiable sentence.
- Line 5: the verification criterion: what you'll count afterwards.
The tool doesn't matter, only that it's always the same place. The value isn't in the single entry, it's in the tenth, when the repetitions surface on their own and you're correcting a tendency rather than a game.
The limit of self-analysis
There's a part of the problem you can't solve alone. The replay shows what you did, not what was available: if you never considered a rotation, that rotation isn't in the recording. The alternative you didn't see stays invisible in review, and watching longer doesn't get you past it.
The second limit is hindsight bias: knowing the outcome, you build a coherent explanation. The death that in real time was a misread becomes "the jungler was invisible, I couldn't have known". Sometimes that's true, often it's the mind protecting itself, and from the inside the two look identical.
So you need two things the replay can't give: an outside eye that sees the options you didn't consider, and a signal that arrives while the decision is still open. The first is a higher-ranked player watching you play, the second a tool running during the game, the reasoning developed in How AI Coaching Helps You Learn League of Legends.
Do one thirty-minute review this week and carry a single correction into the three games that follow; if you want that second eye on while you play rather than the day after, LoL Sensei was built for it.