Methodology

A CPS Test works by timestamping every click with the Performance API and dividing your total click count by the elapsed time in seconds. For timed tests, we anchor the elapsed time to the test's set duration rather than to whatever moment your last click happened to register, so a slow browser tab or a delayed render doesn't distort your score.

Kohi Click Test, Jitter Click Test, and Burst Click Test all use this same underlying measurement. Kohi and Jitter apply it to specific clicking techniques rather than general clicking; Burst Click Test additionally tracks your fastest continuous one-second window within a longer session. Technique-specific breakdowns for these three are still being written; the core timing accuracy above applies to all of them equally in the meantime.

Latency. Keyboard Latency Test splits input delay into two parts. Delivery lag is the time between your keyboard's signal reaching the browser and our JavaScript running in response, driven mostly by your keyboard's USB polling rate and the OS's input scheduling. Processing lag is the time from that JavaScript running to the browser confirming the next frame is ready to draw. This isn't the same as raw USB polling latency, which would need dedicated hardware to measure directly, but it reflects what you actually experience using the keyboard in a browser, and it's a fair way to compare two keyboards or two setups against each other.

Chatter and debounce. Keyboard Double Typing and Mouse Double Click Test both work the same way: every press is timestamped, and if two events from the same key or button land within 80 milliseconds of each other, it's flagged as a hardware fault rather than a real second press. That threshold isn't arbitrary. No human can deliberately press the same key twice in under 80 to 100 milliseconds, so anything faster than that came from the switch itself. This is different from your OS's key repeat feature, which fires repeated events on purpose when you hold a key down; key repeat is expected behavior, chatter is a hardware defect, and the test only counts distinct single presses. When chatter is detected, we look at the worst gap you triggered and suggest a firmware debounce value just above it, capped at 150ms, as a starting point for fixing it in your keyboard's software.

Key rollover (NKRO). Rollover only matters when you're pressing more than one key at once, like an ability key while holding a movement key. Keyboard CPS Test tracks the maximum number of simultaneous keys your keyboard successfully registers. For single-key testing this number doesn't matter, since you're only ever pressing one key; it matters for real gaming scenarios, where a budget keyboard with limited rollover can silently drop inputs when several keys are held together.

CPS Test Kit has three tools that sound similar but measure different things:

Reaction Time Test measures your biological reaction time alone: the gap between a stimulus appearing and you beginning to respond. This is the same phenomenon Franciscus Donders first measured scientifically in 1868, and it typically runs around 220ms under controlled lab conditions, closer to 250 to 280ms in a browser due to added overhead. Simple reaction time (one stimulus, one response) is fastest; recognition (picking the right stimulus among several) adds roughly 100ms; choice reaction time, identifying and selecting among responses, is slower still.

Keyboard Latency Test measures the opposite end: hardware and system delay only, with your biological reaction time deliberately excluded.

Input Latency Test's Clicks Monitor mode is similar to Keyboard Latency Test but for mouse input, isolating hardware and browser processing time. Its Frame-Perfect mode is different: it includes your full reaction loop, biology and hardware together, so its numbers land closer to Reaction Time Test's range (150 to 300ms) than Keyboard Latency Test's.

None of these three measure click-to-photon latency, the true end-to-end delay from a physical button press to light hitting your eyes, which needs dedicated hardware to measure directly. All timing on this site uses performance.now() in the browser, which is precise for comparing your own results over time, but display latency (roughly 1 to 16ms depending on your monitor) sits outside what any of these tools can see. Use them for relative comparison against your own past results, not as an absolute scientific measurement.

Precision Test distinguishes two things people usually conflate: accuracy is how close your clicks land to the target on average; precision is how consistent your clicks are with each other, regardless of whether they're centered. You can be accurate without being precise if your average is good but individual attempts scatter widely, or precise without being accurate if your clicks cluster tightly but consistently off to one side.

You'll also see a Consistency Score on a few different tools, and it isn't computed the same way everywhere. On CPS Test and Typing Speed Test, it's based on how much your timing varies relative to its average, calculated from your click intervals or your second-by-second typing speed. Polling Rate Test uses the same underlying idea but reports it as a letter tier (S+ down to D) instead of a 0 to 100 number. The concept, lower variation meaning more consistent input, is the same across all three; the scale and the underlying data aren't.