What the numbers mean
The Gamepad API normalises every axis to a value between -1 and 1, where 0 is centre. A left stick is two axes: axes[0] is horizontal, axes[1] is vertical. Untouched, both should sit near zero.
Near zero is not exactly zero. Analog sticks are analog: a tiny amount of jitter in the last decimal place is normal and games ignore it with a deadzone. The question is not whether the value moves at all, but whether it settles on an offset and stays there.
The 0.10 threshold this site uses is our own judgement, not a figure from any specification or manufacturer. The Gamepad specification defines the range and says nothing about what counts as at rest. We chose one tenth of full deflection because it sits far above normal jitter and far below a deliberate nudge. Treat it as a line for triage, not a verdict: a stick reading 0.09 that has crept upward over months is heading the same way as one reading 0.12.
Testing it properly
- Put the controller on a flat surface and let go of it. Holding it introduces small deflections you will not notice but the reading will.
- Connect it by USB or Bluetooth and open the gamepad test. Press any button: browsers deliberately hide connected gamepads from a page until a button is pressed, so nothing appears before that.
- Confirm the page reports the standard mapping. If it does not, the button numbers on screen are raw indexes with no reliable meaning, although the axis readings are still usable.
- Watch both sticks for at least ten seconds without touching them. Note the resting value of each axis.
- Now push each stick fully in each direction and let it snap back. A stick that returns to a different resting value than it started from is the clearest single sign of a worn mechanism.
- Save the result or copy the summary. A dated reading is what a warranty claim needs, and you will want the before-and-after if you try a fix.
Why sticks drift
Most controller sticks use potentiometers: a wiper travels across a resistive track, and the resistance at that point becomes the position. The track wears where the wiper spends the most time, which is around the centre, and dust settles in the same place. Both change the resistance the controller reads at rest, and the controller has no way to know its own zero has moved.
That is why drift builds gradually and why it is usually worst on the stick you use most. It also explains why cleaning sometimes helps and sometimes does not: if the cause is contamination, removing it restores the reading, and if the track itself is worn through, nothing short of a new stick module will.
Hall effect and TMR sticks work differently: they read a magnetic field instead of a physical contact, so there is no wiper and no track to wear. Whether that eliminates drift entirely for the life of a controller is a claim about the future, and we have not tested any controller long enough to make it. What can be said is narrower and still useful: the specific wear mechanism that causes most potentiometer drift does not exist in a magnetic sensor.
What to do about it, cheapest first
| Option | When it makes sense | What it costs you |
|---|---|---|
| Recalibrate | Drift appeared suddenly, or after a firmware update | Minutes. Some platforms expose a calibration tool; some controllers reset their centre when you hold them still during a power cycle. |
| Clean the stick | Drift is mild and built up gradually | A can of contact cleaner. Aim into the gap around the stick base, then work the stick through its full range. |
| Warranty claim | The controller is within its warranty period | Your time, plus proof of purchase, the serial number and a dated test result. |
| Replace the stick module | Out of warranty, and you are comfortable opening the controller | A stick module and a soldering iron on most controllers. Drop-in modules exist for some models. |
| Replace the controller | Everything above has failed or does not apply | The price of a controller. If you are buying anyway, a magnetic-sensor stick avoids the wear mechanism described above. |
Compiled by MyGearCheck. The order reflects cost and reversibility, not how likely each one is to work on your specific controller.
Making a warranty claim stick
Support desks see drift claims constantly and triage them on evidence. Three things move a claim along: the serial number, proof of purchase, and a reproducible demonstration of the fault.
The gamepad test gives you the third one in a form that travels. Copy the summary and you get a line of plain text naming the controller, the resting axis values, the date and the page it was measured on. Paste that into the support ticket. It is more specific than a description and it is verifiable by the person reading it, because they can run the same test.
Check the warranty period for your controller with the manufacturer before you write. Periods differ by manufacturer, by region and by whether the controller was bought on its own or in a console bundle, and we are not going to guess yours.
Deadzones are a workaround, not a fix
Many games let you raise the stick deadzone, which tells the game to ignore small deflections. Raising it far enough will hide mild drift.
It costs you precision in exchange. A deadzone large enough to swallow a 0.15 offset also swallows every deliberate small movement, which is most of aiming. It is a reasonable stopgap while you wait for a replacement and a poor permanent answer.
A deadzone also cannot help with drift that has become directional pull rather than a static offset, because the game is ignoring a range, not compensating for a bias.
Frequently asked questions
My controller does not appear in the gamepad test at all. Is it broken?
Almost certainly not. Browsers hide connected gamepads from a page until you press a button on the controller, as a privacy measure against fingerprinting. Press any button and it will appear. If it still does not, try a different cable and a different browser before suspecting the controller.
What resting value counts as drift?
This site treats a consistent reading above roughly 0.10 on an axis as drift. That threshold is our own, chosen because it sits well above normal jitter and well below a deliberate movement. No specification defines one.
Can stick drift fix itself?
No. It can appear to improve after cleaning if the cause was contamination, and it can look different from day to day because temperature affects the reading slightly. Wear on the resistive track does not reverse.
Does recalibrating actually help?
It helps when the controller's idea of centre has shifted but the mechanism is otherwise healthy, which is typical of drift that appeared suddenly. It does not help with a worn track, because there is no stable centre left to calibrate to.
Are Hall effect sticks immune to drift?
They avoid the wear mechanism that causes most drift, because there is no physical contact between a wiper and a track. Calling them immune would be a prediction about years of use that nobody can make from a specification, so treat it as a materially different design rather than a guarantee.
Why does my controller drift only in one game?
Then it is probably not drift. Games apply their own deadzones and response curves, so a controller near the threshold can be visibly off in a game with a small deadzone and fine in one with a large deadzone. Confirm with the browser test, which applies no deadzone at all.
Does the gamepad test work with a controller connected by Bluetooth?
Yes. The Gamepad API does not distinguish between USB and Bluetooth, so a paired controller behaves the same. If a Bluetooth controller reads worse than the same controller on a cable, suspect the connection rather than the stick.
Can the browser tell me which stick component has failed?
No. It reports the values the controller sends and nothing about what produced them. Whether the cause is a worn potentiometer, contamination or a calibration offset is a question you answer by trying the fixes in order, not one the test can answer.