How much stick drift is normal?
A perfectly still stick almost never reads exactly zero. The useful question is not whether your controller moves at rest, but how far it moves, whether it leans in one direction, and whether the result repeats.
Noise versus drift
Sensor noise wanders. The reported position bounces either side of centre by a few thousandths and averages out to roughly nothing. Every analog stick produces some variation, and a game's inner deadzone normally absorbs small changes.
Drift leans. The reported position sits away from centre and stays there, so the game receives a constant small push in one direction. That repeatable directional offset is the pattern worth measuring.
A stick can also return to different positions after being moved. That recentring problem may be missed by a single resting sample, so repeat the test after rotating the stick and letting it snap back to centre.
How GamePadTestLab labels resting values
GamePadTestLab evaluates the radial distance from the browser-reported centre and uses the 95th-percentile resting offset for its verdict. The 95th percentile describes the typical high end of the sample without letting one isolated spike decide the result.
These bands are application-defined guidance for browser-visible readings. They are not universal industry limits, manufacturer calibration specifications, or warranty thresholds.
| 95th-percentile offset | Label | Practical interpretation |
|---|---|---|
| 0 to 0.01 | Healthy | Within the site's default sensor-noise range for this sample |
| Above 0.01 to 0.04 | Minor deviation | Small repeatable movement that many game deadzones may hide |
| Above 0.04 to 0.09 | Potential stick drift | Meaningful resting movement worth retesting and comparing in a game |
| Above 0.09 | Significant stick drift | Large browser-visible movement that is more likely to affect play |
Run a repeatable stick drift test
Temperature, a recently moved stick, the surface under the controller, and software between the controller and browser can all shift one sample. A repeatable procedure is more useful than one screenshot.
- Connect the controller, focus the page, and press a button so the browser exposes it.
- Place the controller on a stable surface and make sure nothing touches either stick.
- Open the stick drift test and run the full nine-second resting measurement.
- Compare the left and right 95th-percentile offsets, centre coordinates, and jitter rather than relying only on the verdict label.
- Rotate each stick once around its edge, release it, and run the measurement again.
- Repeat once more over the same connection. A similar directional offset across runs is more persuasive than a single spike.
Examples of how to read a result
Illustrative example: a stick that fluctuates around the centre but has mean X and Y values close to zero is showing noise rather than a consistent directional lean. These are example patterns, not measurements taken from a particular controller.
Illustrative example: a right stick that repeatedly settles to one side after release, while its 95th-percentile offset stays above the site's potential-drift boundary, deserves comparison in a slow camera movement test. The browser result supports the diagnosis but does not identify the physical cause.
Illustrative example: one unusually high peak with a much lower median and 95th percentile may be a brief bump or scheduling irregularity. Retest before treating the peak as damage.
What to do next
If the offset is repeatable, use the controller deadzone test to find the smallest inner deadzone that contains the resting movement. A modest adjustment may make a controller comfortable to use, but it trades away some fine movement near the centre.
A deadzone is a workaround, not a repair. If the required setting noticeably harms aiming, the value changes sharply between runs, or the stick cannot return consistently, cleaning, professional repair, a warranty claim, or replacement may be more appropriate.
Limits of a browser result
The Gamepad API reports values after the controller firmware, operating system, driver, connection layer, and browser have handled them. Built-in calibration or remapping software may therefore change what this page sees.
GamePadTestLab cannot inspect the stick mechanism, identify dust or worn parts, recalibrate firmware, or decide warranty eligibility. Use repeated browser readings together with real in-game behaviour and manufacturer support guidance.
Ready to check your own controller?
Run the stick drift test