A mouse rarely dies all at once. It develops one specific fault while everything else keeps working, which is why "my mouse is acting up" is so hard to act on. Four checks, each taking under a minute in a browser tab, separate a failing button from a failing cable, a failing hand, or a setting. Run them in this order, because each one rules out what the next would otherwise confuse.
First, rule out the computer
Before testing anything, plug the mouse into a different computer, or plug a different mouse into this one. It takes a minute and it is the only step that cleanly separates the device from everything around it. Skipping it is how people end up replacing a perfectly good mouse because a USB port had gone bad.
If it is wireless, put in a fresh battery first. A dying cell produces symptoms that look exactly like several hardware faults at once: skipped clicks, cursor jumps, and dropouts that come and go with distance.
Test one: does every button report at all
Start with presence, not quality. Left, right, middle, and both side buttons should each register something. Press each one several times and watch for the button that does nothing, or the one that needs a harder press than its neighbours.
A button that works only when pressed hard is usually not the switch. It is more often the plastic shell flexing or a screw that has worked loose, and tightening the case can bring it straight back. A button that never reports at all, on two different computers, is a broken switch or a broken solder joint.
Test two: is one click arriving as two
This is the most common failure on a mouse older than a year or two, and it is the one where a measurement beats an impression. Use the Double Click Test, which logs the gap between consecutive clicks in milliseconds.
Read it by the numbers rather than the feel. Deliberate double-clicks from a human hand land somewhere in the 100 ms to 400 ms range. A gap under about 40 ms is faster than anyone can click on purpose, so a cluster down there means the hardware sent a click you did not make. Click slowly and single-click only for thirty seconds; anything the log records is not yours.
Test three: does the button hold
A worn switch can fail in the opposite direction too. Instead of adding a click it drops one, releasing in the middle of a press. You notice it as a drag that lets go halfway, a selection that collapses, or a weapon that stops firing while you are still holding the button.
The Mouse Drag Test holds a press and shows whether it stays held. Press, keep holding, and move slowly for ten seconds. A clean mouse shows one unbroken hold. A failing one shows the press breaking into pieces. This test matters because the drop and the double-click have the same underlying cause, so a mouse that passes test two can still fail here.
Test four: establish what normal looks like for you
The last check is a baseline rather than a fault test. Run the Click Speed Test and note the number. Most people land around 6 to 7 clicks per second. It is worth recording while the mouse is behaving, because a noticeable drop later is a useful early signal that a switch is stiffening before it starts misfiring.
It also settles an argument with yourself. If clicks feel like they are being missed but the rate is unchanged, the problem is more likely perception or an application than the mouse.
Reading the results together
Take the four outcomes as a set. A button that never reports is electrical. A button that reports twice is a worn contact. A button that will not hold is the same wear showing a different face. A mouse that passes all four while still feeling wrong is usually pointing at something outside the device: a surface the sensor dislikes, a polling or acceleration setting, or a utility from the manufacturer running in the background.
If test two is what failed, the causes and the repair options are a separate read. If two clicks only ever happen when you are deliberately clicking two things in a row, nothing is broken and the double-click speed setting is the control you actually want.
What a browser test cannot tell you
Being straight about the ceiling here matters more than the tests. A page in a browser sees the events the operating system passes it, and nothing underneath that. It cannot distinguish a worn switch from a driver or a manufacturer utility generating clicks on its own, because both arrive looking identical. It cannot read the sensor, measure polling rate at the hardware level, or see anything about the cable short of the dropouts it causes.
That is why the first step in this guide is swapping the mouse to another machine. It is the one test that reaches past what a browser can observe, and it costs a minute.