How AimBench works

What the Setup Score measures

How well your hardware is matched for competitive play, on a 0–100 scale: 0 is a stock office build (60 Hz office monitor, 125 Hz office mouse, membrane keyboard, bundled cloth pad; a stock headset off onboard audio if you add audio), and 100 is every domain at the point where more gear stops measurably helping. It is not a rank, a skill rating or a win-rate prediction. It's built from the same flags the diagnostics list, so the score and the bottleneck list can't disagree.

Six domains, weighted by the evidence

Each domain takes its worst flag (the limiting factor) and turns it into points: good 100, info 80, warning 50, critical 20. The headline is the weighted average, remapped so the office build reads 0. Without audio, the four aim domains are re-weighted to fill the whole score (they carry 68% of it when audio is on).

Curves above the floor

A step score can't tell a 144 Hz panel from a 540 Hz one. So past the floor, Display follows frame time and the input chain follows the polling interval, both in a straight line, flattening where the evidence says the gain stops (360 Hz for refresh, 2000 Hz for polling). Below the floor the warning points apply.

Controller mode

A pad has no mouse chain, surface or keyboard, so one Controller domain takes their 44% share and display and audio weigh the same as they do for mouse players. It scores two things. Measured input latency on the connection you use, averaged by an independent tester (a maker's figure never fills the slot, and an unmeasured connection is left out rather than guessed): 2 ms or less scores 100, falling in a straight line to 80 at one 144 Hz frame (6.9 ms); more than a frame is a warning and more than two is critical. Back paddles: without them your right thumb leaves the aim stick every time you jump, crouch or slide, so no paddles is a warning. Stick type (potentiometer sticks drift as they wear) and your deadzone are shown as advice. The 0 point is a stock first-party pad over Bluetooth on a 60 Hz panel.

Some flags are shown but never scored: whether your pad's glide suits your aim style, your sensitivity band, FPS under your refresh rate, mouse weight. They're choices or free settings, not gear quality. Scoring FPS once made a 240 Hz monitor score lower than a 144 Hz one at the same frame rate, which is exactly backwards.

The evidence ladder

The weights and the upgrade order follow how big each effect actually is, from the research summarised in Does gear give a competitive advantage? The short version: gear is a floor, not a booster. Below the floor it holds you back; above it, spending buys feel and consistency.

“Baseline cleared”

When no domain carries a warning or a critical flag, the dashboard says your setup has cleared the competitive baseline. Info flags (small, real headroom such as 144 → 240 Hz) don't block it. That's the honest point to stop upgrading and practise.

Upgrade picks

For each fixable flag the engine swaps every product in that category into your setup, re-scores it, and keeps only the ones that raise your score, best per price class. Ties break on the measured build score, then on price, never on catalog order or popularity. Prices are shown as a class (Budget to Flagship), never a number: your local store has the live price.

Sensitivity and cm/360

cm/360 is how far your mouse travels for one full turn: (360 × 2.54) / (DPI × sens × yaw), where yaw is each game's degrees-per-count constant. It doesn't depend on the game, which is why it can compare players across games and convert a setting from one to another. WARDOGS is the exception to the simple formula: its slider is affine (it has an offset), so it's converted with its own two constants. Each constant was checked against a published reference and against pros' own settings; games whose scale isn't a clean constant are marked approximate.

Where the pro data comes from

215 tracked players, each with a source link and a check date. The player's own word wins: a chat command, a channel panel or their own video beats an aggregator, and 79 records come straight from one. Values are never inferred. If a source swaps a named product for something unidentifiable (“Custom keyboard”), the field is dropped rather than guessed, and an aggregator whose page hasn't changed in years is dated to when it last changed, not to when we last read it.

Where the specs come from

Hard specs come from the manufacturer's page, with independent measurements where a spec sheet hides the real difference: click latency and sensor behaviour from TechPowerUp, monitor response from the maker's published figures, headphone and amp data from RTINGS, Audio Science Review and Crinacle. Friction (glide), drive-match and mic gain are models built on those specs, labelled as models wherever they appear. Mousepad and skate build grades are reviewer consensus, cited per product.

What it can't tell you

Whether a shape fits your hand (that's the hand-size finder's job), how a switch feels, or whether you'll get better. Above the floor, practice beats gear by a distance.