A demo review that takes 20 minutes and actually finds something

Most demo review fails for the same reason: people watch the whole match from their own POV, wince at the deaths they already remember, and finish with a vague sense that they should play better.

That produces nothing. A useful review answers one question and ends with one change.

Before you open anything

Pick the question first. Not “how did I play” — something like:

• Why is my opening-duel record negative?
• Where does my utility go?
• What happens in the three rounds after I lose an eco?

One question per demo. If you do not have one, get it from your stats rather than your memory: your memory keeps the highlights and the humiliations, and forgets the pattern.

The 20-minute pass

Minutes 0–5: the first-contact pass.

Skip to the first fifteen seconds of each round and stop. Only note two things: were you the first to take a duel, and did you have information before you took it?

Do not watch the rest of the round. You are counting a behaviour, not analysing outcomes.

Fifteen rounds is enough. If you were first contact in more than three or four of them and you are not the designated entry, you have found something.

Minutes 5–10: the buy pass.

Freeze-time only. For each round write down: money, what you bought, and what the loss bonus was. This is tedious and it is the highest-yield five minutes in the exercise, because economy errors are invisible in the moment and obvious in a list.

Look for the round where you bought at $2,300 and the two rounds after it.

Minutes 10–15: the utility pass.

Watch only your grenades. For each: what did it accomplish, and did anyone use it? A flash nobody followed is $200 spent on a warning. If more than half your flashes fall in that bucket, that is the finding.

Minutes 15–20: the death pass.

Now watch your deaths, but from the enemy’s POV, not yours. This is the single most valuable switch in demo review. From your own POV a death looks like a lost duel. From theirs it usually looks like you walked into a place they were already looking, which is a different problem with a different fix.

A worked pass, start to finish

The framework is easier to trust with an example attached. Take a real profile from our worked examples: Level 8, 1,279 matches, 51% win rate, and a report whose top finding is utility output at 3.5 damage per round against a 4.57 median for their level.

The question: where does my utility go?

The pass: watch only their grenades across fifteen rounds. Two things are already known from the stats before the demo opens — they throw 9,701 flashes against 6,305 non-flash grenades, so their mix is fine, and their flash success rate is 46%, which is normal. So the demo is not looking for “do they buy the wrong things”. It is looking for placement and timing.

What that narrows it to: grenades thrown at the wrong moment (before anyone is in position to use the space), or at the right moment into the wrong place (a molotov that lands where the enemy already is not).

The sentence: “When we take B, I throw my molotov as I arrive. Instead I will throw it ten seconds before the take, so the angle is already denied when we cross.”

That is a review that could not have been produced by watching the whole match and wincing. The stats did the narrowing; the demo did the diagnosis; the sentence did the commitment.

Write one sentence

End every review with one sentence in this form:

In situation X, I do Y. Instead I will do Z.

For example: “When we take B, I am the first through the smoke. Instead I will wait for the first contact and enter behind it.”

If you cannot write the sentence, the review found nothing and that is fine — pick a different question next time.

The questions worth asking, in rough order

If you do not have a question, these are the ones that most often produce a sentence. Take them in order and stop at the first one where the answer surprises you.

1. Was I first contact, and did I know anything before I took the fight? The highest-yield question at every level, and the one most likely to overturn what you believe about your own play.
2. Where was I standing when the first person on my team died? Not what you did next — where you were. Six rounds of this usually settles whether you are trading or spectating.
3. What did each of my grenades accomplish, and did anyone use it? Ruthlessly: a flash nobody followed accomplished nothing.
4. What did I buy, and what rung was the ladder on? Tedious, and it finds the errors that are invisible in real time.
5. In my deaths, what did the enemy see? Watch from their POV. From yours it is a lost duel; from theirs it is usually someone walking into a place they were already looking.

Questions 1 and 2 are the same finding from two angles, which is why they sit at the top. If both come back clean, your positioning is fine and the answer is further down.

What not to do

Do not review a match you lost badly. You will find ten things, fix none, and feel worse. Review a close one, or better, a win — leaks in a won match are the ones you would otherwise never notice.

Do not review at 1am after the session. You are reviewing tired play while tired.

Do not watch the whole thing. Forty minutes of demo produces less than twenty structured minutes.

Do not collect findings. One change at a time, held for thirty matches. Three simultaneous changes cannot be evaluated, so you learn nothing from any of them.

Two passes worth adding once the basics are habit

The trade pass. Pick ten rounds and, for each one, note where you were standing when the first member of your team died. Not what you did next — where you were. Most players discover they were somewhere unrelated to the fight in six or seven of the ten, which is a positioning finding rather than a duelling one, and it is invisible from any other angle.

The session pass. Not a demo at all. Take your last thirty matches, group them wherever there is a gap of more than about three hours, and write the net result of each sitting. If the long sessions are the negative ones, no amount of demo review will fix what you are looking at — the leak is in the calendar, and one of our worked examples shows exactly that shape: 54% headshots, a positive opening duel balance across 11,085 duels, and two evenings that cost 225 ELO between them.

Worth running the session pass first, actually. It takes five minutes and it occasionally makes the other twenty unnecessary.

How often

One demo a week is plenty if each produces one sentence and you actually apply it. Reviewing three a week and changing nothing is theatre.

The measurement loop is: pick a leak, write the sentence, play thirty matches, check whether the number moved. If it did not move, either the change did not happen or it was not the leak — and both of those are useful to know.

Getting the question

The fastest way to a good question is to look at what your own numbers already say. Run your profile and the top-ranked leak is the question for your next demo.