How to read your controller drift test
A stick sending movement while resting produces an offset from the centre. Driftbench measures that input for ten seconds. The result describes the signal reaching your browser; it does not, by itself, confirm a faulty sensor.
Make your test repeatable
- Connect the controller, press a button and check that the sticks respond.
- Place it on a stable surface. Start the test and release the sticks during the three-second countdown.
- Leave it untouched during measurement. Download the report and repeat with the same connection, browser and settings.
What each number means
- Mean: average distance from the centre across all samples. It summarises offset throughout the test.
- Peak: largest observed distance. It can come from a brief movement; compare it with the mean.
- P95: distance at or below which at least 95% of samples fall. It helps separate typical behaviour from the most extreme values.
A percentage example
Mathematical example, not a physical controller test: if X = 0.03 and Y = 0.04, distance from the centre is 0.05, displayed as 5%. 100% represents full travel along one axis. A diagonal can exceed 100%; it is not a percentage of damage.
How much drift is acceptable?
There is no universal threshold. Compare the offset with behaviour in your game and its deadzone. Driftbench’s deadzone control only draws a reference: it does not calibrate the controller or change game settings. A zero reading does not prove that a sensor is unworn.
Compare before and after
Save the initial report, note your connection and settings, then repeat after a change. Compare both sticks, mean, peak and P95. Repeat several times before attributing a difference to the change: touching the controller, switching connections or using another browser can affect the comparison.
Try it with your controller
Open the controller drift test →
Controller not detected? What to check