How we measure

When you upload a match replay (your .dem file), NextFrag plays it back and measures what the replay records: where you aimed, when you fired, what you hit, and how you moved.

Your three scores: Aim, Movement, Overall
Each is a number from 0 to 100 on a fixed reference scale. Higher is better. 100 marks the top end of the community reference this scale was fitted to. It is not a perfect score, and it is not a promise about the matches you will win. The reference stays fixed so two matches can be compared. A later scoring update can move a number even when the play did not change.

Aim — first-bullet placement, recoil and spray control, firing discipline, line-of-sight reaction, crosshair movement, and weapon handling. When both readings are available, one fifth of Aim is the hitting reading and the rest is the mechanics reading. If one reading is missing, Aim uses the one that is there.
Movement — stopping before you shoot, changing direction, peeking, time exposed, how you move when you are not shooting, and how quickly you reposition after a rifle burst that does not get a kill.
Overall — your single all-round score, calculated on its own. It is not an average of Aim and Movement. It uses many of the same measurements, plus how far away your fights are, whether you take them alone, damage you deal before you die, and how fast you are moving when you throw a grenade. This is the score on the leaderboard.

Aim, Movement and Overall each appear on the analyzed match when there's enough data. A measurement that is missing is left out. It is not filled in as zero. When a score cannot be worked out, the page shows a dash (—).

What the comparison measures
In the reference set this scale was fitted on, the score ranks the pro higher this often, counting ties as half. The counts are player-match records, one player in one match. They are not a count of demo files, and they are not the live leaderboard.

Aim: 88.7%, on 1,907 pro and 1,192 community player-matches.
Movement: 97.8%, on 1,905 pro and 1,192 community player-matches.
Overall: 98.6%, on 1,907 pro and 1,192 community player-matches.

All three use 185 pro players and 961 community players. Separating professional matches from community matches does not say how well a score orders players inside the community. These scores are not a Premier rating, a FACEIT Elo, or a match win rate. Opponents, teammates, and the fights you take still affect some inputs, especially Overall.

How sure the read is
With no extra credit for hits, 99 recorded shots reaches a Decent read and 388 reaches a Reliable read. Enough hits can reach those labels with fewer shots. Below a Decent read the page says Light read and adds an early-signal note. The label does not change your score, and it does not require a minimum number of rounds. A Decent read or a Reliable read describes the shot sample. It is a different label from the band beside the number.

On the anonymous first report, Overall shows a band beside the number: Beginner, Developing, Solid, Advanced or Elite. That word is a label for the same 0 to 100 number, not a second score. Signed-in match pages show the number without that band.

Reaction
The reaction inside Aim, and the Reaction time number on the match page, is line-of-sight reaction: the time from an enemy becoming visible to your first aimed shot. A deeper section can also show return-shot latency, which starts when you are hit, and a view-cone reaction, which starts when an enemy enters your view cone.

Coaching on the match
Depending on how much you played, the match page adds coaching on the habits behind the scores, and some measurements sit next to numbers from professional matches run through the same analysis. Professional matches are tougher, so treat the comparison as a guide.

When the anonymous first report has a leading individual issue, it is labelled Fix this first. When the signed-in report shows a leading finding, it is labelled Today's focus. It is the highest-priority individual practice focus, not a promise of the biggest score increase or of more wins. Where a finding includes them, you also get a drill, a target, and a way to check again. Repeated advice is limited, so related habits do not each get their own copy of the same drill. When no individual issue clears the bar, the report says so and does not invent a drill. Sensitivity advice appears only when the match has enough flick shots.

Getting onto the leaderboard
To appear you need at least 3 competitive matches on the current scoring where we could work out all three scores, a display name, and to opt in. Only competitive matches count. Deathmatch does not. When the leaderboard shows a score, it combines those competitive matches, and a match analyzed 60 days ago counts half as much as one analyzed today. Staying off the board does not stop the scores being calculated. Joining the public ranking is opt-in.

Your data
Your demo file is deleted after a limited retention period. We keep your analysis results — your scores and the coaching breakdowns — in your account history until you delete the record or your account. Read the Privacy Policy for the detail.

▼ Scoring updates

9 June 2026 — scoring update. All three scores were rebuilt. Inputs that repeated another input, and inputs that tracked shot volume or raw accuracy rather than the mechanic, were removed.
17–21 June 2026 — scoring update. The share of Aim that comes from the hitting reading moved from 35% to 20%, which is still the share today. Weapon handling, time exposed, and a line-of-sight crosshair measure were added.
5 July 2026 — scoring update. Outcome, volume, role, and round-phase inputs were removed from the scores, along with inputs from an April movement model.
16 July 2026 — scoring update. Firing too late after stopping was brought into Movement and Overall.
9 September 2026 — scoring update. A crosshair-path measure was restored to Aim only. It does not move Overall.
19 September 2026 — scoring update. Inputs that were inconsistent from match to match were given less influence on Movement and Overall. Aim was left unchanged that day.
23 October 2026 — scoring update, live since 24 October 2026. Movement and Overall now include how quickly you reposition after a rifle burst that does not get the kill, and the weights for Aim, Movement and Overall were refitted.
1 November 2026 — scoring update. New uploads now use every input the Aim, Movement and Overall scales were built on. Earlier uploads were missing three of them, so their scores could differ by several points. Progress, personal bests and goals only compare matches measured the same way, so existing score goals need their targets confirmed or changed.

Where we're upfront about limits: A replay shows what the game server saw — not your monitor, mouse, or connection — so things like input lag aren't in it. One crosshair-path input inside Aim can vary with your hardware. The measurements behind this scale were checked on server-recorded GOTV demos. That check does not establish the same result for a demo recorded from your own point of view. Professional comparisons come from tougher matches, so they are a guide, not an exact match.

Stop guessing why rounds slip. Find the exact mechanic leaking kills.

NextFrag analyzes your CS2 demo tick-by-tick and ranks your biggest leaks in a few minutes — so your next practice block has one clear priority.

No demo on this device? See a full example report

Free: 1 analysis a day. No account needed.

Analyze your own point of view in this demo.

Drop your CS2 .dem file here

or click to browse — .dem, .dem.zst (FACEIT) or .dem.bz2 (Premier) · no Steam link required for manual upload

Where is my demo?

Demo files deleted after a limited retention period · How we measure

Your demo is on your gaming PC. Open nextfrag.gg there, or read the example report.

Anonymous uploads: max 350 MB. Sign in to save history; upgrade to Pro for larger uploads.

How to get your CS2 demo

Download your match in CS2, then upload the .dem file here.

Open CS2

Go to Watch → Your Matches.
Download the match

Click the download arrow icon (↓) on the match row. Wait for it to finish — it turns into a Watch button when done.
Find the file

Windows default:

C:\Program Files (x86)\Steam\steamapps\common\Counter-Strike Global Offensive\game\csgo\replays\

Steam installed elsewhere? Steam → Counter-Strike 2 → Properties → Installed Files → Browse, then open game → csgo → replays.
Upload it

Drag the file straight into NextFrag — no need to unzip it. Valve Premier gives a .dem.bz2; FACEIT gives a .dem.zst — both work as-is.

Tip — repeat uploads

Create a desktop shortcut to your replays folder so you do not have to browse there every time.

FACEIT demos

Download FACEIT demos from the matchroom and upload them directly — .dem.zst works without extracting. Valve Premier .dem.bz2 files work too. Available account connections are shown in Control Center after sign-up.

Got the file? Drop it in the upload box above. ↑

New Team demo analysis

Set up auto-import (1 min)

Opens the import setup. You can connect Valve match history or FACEIT, then keep uploading demos yourself.

Ranking

Find the mistake you repeat. Train it. Check it on your next demo.

1 · Find the leak

Your top 3 leaks, ranked by practice priority.

Counter-strafe: stop before you shoot · #1

2 · See why

See which mechanics are holding your score back.

Held back by counter-strafing and crosshair pre-aim

3 · Train it

Trainable leaks come with a routine, setup command, or review round.

Train this · Open practice map · Review round 8 — example

4 · Re-measure

Play 3 matches with the drill, then upload them. One match can swing a lot; three help show a pattern.

A+D key overlap · example readings: 41 ms and 24 ms

Upload a demo manually. Available match connections are shown in your Control Center.

Proof, not promises

Privacy

Demo files are auto-deleted

Uploads are used for analysis, then removed automatically after a limited retention period. You can run a first analysis without linking Steam.

Reliability

Confidence is shown when data is weak

Each analysis shows its reliability. A low shot count or weak signal is flagged, not hidden.

Method

Pro benchmarks use the same analysis

Pro demos and your demos run through the exact same pipeline, so the comparison stays honest.

No mockups — our examples are real demos

Examples

Any sample analysis we feature comes from an actual demo parse — direct metric evidence and coaching, not a mockup.

Open sample analyses →

What we measure — and how confident we are

Every metric is classified by data source.

Directly measured

Parsed tick-by-tick from your .dem file.

Calculated from your demo

Formulas applied to those measured inputs.

Estimated / Experimental

Clearly labeled. It never counts toward your Aim score.

Pro benchmarks come from real pro match demos through the same pipeline (shown with sample size); every analysis carries a sample reliability score and flags low sample counts — we don't hide uncertainty.

Full methodology — formulas, weights, limitations → What demo analysis can and cannot measure

Free vs Pro

Start free to diagnose one demo per day. Pro shows every finding for each demo and up to 200 matches. With 6 competitive matches in 30 days, compare your last 3 with the 3 before against measured noise bands.

	Basic (free)	Pro

Analyzing your demo

Keep this tab open until your report is ready.

While we analyze: a free account keeps this report.

Select player

Add your sensitivity (optional)

Sensitivity

DPI

cm/360 (auto)

Sensitivity and DPI are optional — they only sharpen the sensitivity recommendation. The rest of the analysis runs without them.

Privacy Policy

Impressum

How we measure

Every finding for each demo, and your matches over time.

Sign in

Reset password

Control Center

Delete account

Hey!