Click Speed Test Strategy: What CPS, Burst and Spread Actually Measure (2026)

1. What a fixed-window click test actually measures

There are two ways to time a click test, and they do not agree. One is to start a stopwatch, let the player click, and stop when the player stops. The other is to declare the window first and count presses inside it. Click Sprint is the second kind: you pick 1, 5, 10 or 60 seconds, the run lasts exactly that long, and clicks per second = counted clicks ÷ the full window. A run that ends early is not faster; it is just a shorter run with fewer clicks. That single decision is why the score on this page never depends on when you choose to stop.

The default mode is 10 seconds with the left mouse button. The chips change the length and the input; they are disabled while a run is in progress, so one run is always one clean measurement. The URL can pin a mode, for example ?duration=5&button=space, which is useful for practising the same thing twice.

Four numbers come out of one run:

The fifth label, the pace band, is editorial: Steady below 3 CPS, Brisk from 3, Fast from 5, Very fast from 7, Blazing from 10. It describes your own number. It is not a percentile, because the page has measured no other players and makes no population claim. If you want a target, use the arithmetic in section 5 rather than an "average player" figure you cannot verify.

2. The clock starts on press one — so there is no warm-up press

The first press does three things at once: it starts the clock, it counts as click one, and it sets the timestamp every later gap is measured from. The run ends when a press arrives at or after the window, or when the animation loop notices the window has passed; either way, the window is fixed.

That has a practical consequence that catches people on their first attempt. You cannot tap the pad once to "get ready" and then start clicking in earnest: that settling tap is the run. If you spend the next 300 ms finding a comfortable grip, those 300 ms are inside the window and the clicks you did not make are missing from the count. On a 1-second run, 300 ms of settling is almost a third of the test.

The correct sequence is therefore:

1. Pick the length and the input while the page is idle.
2. Put your hand in the position you will finish in — the same grip, the same wrist angle, the same distance from the pad.
3. Make the first press the first click you actually want counted.

There is no penalty for taking your time before that first press. The clock is not running while the page shows "Tap to start". Use that: set your hand, breathe once, then go. What you must not do is start the clock with a throwaway press.

The same rule explains why "Go again" exists. It returns the page to the ready state without starting anything, so the next run begins on your next press, not on the button that reset the previous one.

3. The 12 ms gate: an interval floor, not a human detector

Every press must be at least 12 milliseconds after the last press that was counted. Anything closer is dropped and the result reports how many presses were dropped. The reason is physical: USB polling runs at roughly 1–8 ms, but a hand on a real switch cannot close it twice inside 12 ms. A timestamp pair closer than that is a synthesised event, a double-fire from the hardware, or an autoclicker.

What the gate does well is make the report honest. A naive autoclicker firing every 1 ms does not post 1,000 CPS; it posts a number near the interval floor and says how many presses it threw away. I ran perfect synthetic streams through the shipped classify function to see the exact shape:

Read the table carefully, because it says two different things. First, the ceiling for a perfectly spaced synthetic stream is about 83–84 CPS, not thousands. Second, the gate is an interval floor, not a fraud detector: a clicker paced at 5 ms still gets 67 CPS through, because every third press is more than 12 ms after the last accepted one. The page never claims to detect software; it claims to refuse intervals a physical button cannot produce, and it reports the refusals instead of hiding them.

For a human run the gate should never fire. Even a very fast tap is far slower than one press every 12 ms — 12 ms would be 83 clicks in a second. If you see dropped presses in your result, something is double-firing: a mouse with a failing switch, a macro, or a touchscreen generating duplicate pointer events. That is useful diagnostic information, and it is why the count is shown instead of silently discarded.

One more detail that matters for the 60-second run: the gate compares each press with the last accepted press, not with the previous raw event. A burst of impossible events therefore cannot "use up" the gate and block the next real press; the next real press is measured against the last press that actually counted.

4. The four numbers describe four different runs

Two players can both report 5.0 CPS and have completely different runs. The average is a single division; the burst and the spread describe the shape that the average hides. Here are three constructed 5-second runs, each with exactly 25 clicks, so each reports exactly 5.0 CPS, 200 ms pace. I built the timestamp lists by hand and pushed them through the shipped summarise; they are worked examples, not player data.

All three are the same score. Only one of them is a rhythm.

• Average CPS is the number the page is named after, and the least diagnostic. It is the right number for comparing two runs of the same length, and the wrong number for understanding either one.
• One-second burst is the fastest second anywhere in the run. In a 1-second run it equals the click count by definition. In a 60-second run it is usually much higher than the average, because it is a peak, not a pace. A high burst with a low average means you can produce the speed but cannot hold it.
• Pace in milliseconds is the actionable version of CPS. "5 CPS" is an abstraction; "200 ms per click" is a target you can feel. To improve, you are always trying to remove milliseconds from this number.
• Spread is the population standard deviation of the gaps between counted presses. Zero means metronome-perfect; 82 ms means the gaps are all over the place. At the same average, a lower spread is the better run, because the fast clicks in a spiky run are being paid for by dead time you could have used.

The four numbers also have different noise floors. A single missed click changes a 1-second score by 1.0 CPS, a 5-second score by 0.2, a 10-second score by 0.1 and a 60-second score by about 0.017. Short runs are sharp instruments and unstable measurements; long runs are stable and blurry.

5. The four windows are four different tests

The length is not a preference. It changes which physical quality dominates.

The arithmetic of a pause makes the difference concrete. Suppose your natural pace is 6 CPS and you stop for one full second somewhere in the run:

In a 1-second run the same stop leaves nothing at all. This is why a 1-second personal best and a 60-second personal best are not the same skill, and why a 60-second average that looks modest can hide a genuinely fast first ten seconds. Here is that shape as a constructed example: 80 clicks at 8 CPS for the first 10 seconds, then 250 clicks at 5 CPS for the remaining 50. The result is 330 clicks, 5.50 CPS, burst 8, pace 182 ms. The average is 5.50 even though half the run was much faster.

Practical reading:

• 1 second is a start-up test. The score is dominated by how quickly you get the first two or three clicks down. One slow start and the run is over.
• 5 seconds is long enough to show a rhythm and short enough that a single stall is expensive. It is the best length for practising a new input.
• 10 seconds is the default for a reason: long enough that pacing matters, short enough that fatigue is not yet the story. Most practice should happen here.
• 60 seconds is an endurance test. Do not judge your top speed from it; judge how little the average falls away. The burst stays high while the average decays — that gap is the fatigue signal.

6. Inputs: what the page timestamps, and why travel matters

The page timestamps the press, not the release: pointerdown for mouse and touch, keydown for the keyboard. That is the single most useful mechanical fact in this guide. The time your finger spends coming back up is not measured; the time it spends going down is. You need enough release travel to re-actuate the switch, and no more.

The three inputs are not three skins on the same event:

• Left click. The pad listens for pointerdown with the primary button. A mouse switch has a short travel and a tactile or audible actuation point; the event fires when the button actuates, before you bottom out against the shell. Pressing past actuation adds force and noise, not speed. The practical rule is to press until it clicks and stop, rather than driving the button to the end of its travel.
• Right click. Identical maths, different finger and a different switch on most mice. The page suppresses the context menu on the pad, so the right button can be used as a pure click. Two cautions: many mice have a stiffer right switch, and a touchscreen cannot produce a right button at all — pointerdown from a finger carries the primary button, so right-click mode is a mouse mode.
• Spacebar. This is a keyboard switch with much longer travel than a mouse button, and the page ignores key repeat, so holding the key down does nothing after the first event. One press is one click, and the key has to come back up past its reset point before the next one counts. That reset travel is the cost of this input. Space mode also ignores pointer presses on the pad, so on a phone without a keyboard the space mode has no usable input; use left click there.

Two smaller facts complete the picture. With the pad focused, Enter or Space counts as a press in any input mode, which makes keyboard testing easy. And global key handling deliberately ignores keys aimed at buttons, links or form fields, so tabbing to "Rules" and pressing Space opens the rules instead of counting a click.

Which input is fastest is not something this page can tell you, and it does not pretend to: it measures the timestamps you produce, not the hardware. What it can tell you is which input you are personally more consistent on. Run the same 10-second test on each of the three, twice, and compare spread before CPS. The input with the lower spread is the one to practise; speed follows consistency, not the other way round.

7. Why rhythm beats force

At 5 CPS you have 200 ms per click. At 6 CPS you have 167 ms. At 7, 143 ms. The gaps shrink fast at the bottom and slowly at the top:

This is why "click harder" stops working. Force does not shorten the switch's actuation travel; it only makes your hand tired and your gaps uneven. The improvements that remain at the top of the table are small timing corrections, and small timing corrections are exactly what rhythm practice trains.

Spread is the trainable part. A run with a low spread at 5 CPS can be pushed to 6 CPS by shaving a few milliseconds from every gap; a run with ±80 ms spread cannot, because its average is being held down by stalls rather than by a uniform ceiling. If your burst is much higher than your average, you already have the speed — you are losing it between bursts. If your burst is close to your average and both are low, you are at your current ceiling and need a mechanical change (a different finger, a different input, a shorter travel), not more effort.

There is one more reason to prefer rhythm: it is repeatable. A lucky fast run is not a skill. A 5.2 CPS run with a small spread is something you can reproduce tomorrow, and reproducibility is what makes practice accumulate.

8. Reading your own result card

The result card gives you five pieces of evidence. Read them in this order:

1. Clicks and CPS. Did the run land where you expected? On a 10-second run, each click is 0.1 CPS, so a target of 5.0 CPS means 50 clicks.
2. Burst. If the burst is far above the CPS, the speed exists but the pacing does not. If it is close, the run was uniform — good or bad depending on the average.
3. Pace. Convert the CPS you want into milliseconds before you start: 5 CPS = 200 ms, 6 = 167 ms, 7 = 143 ms, 8 = 125 ms, 9 = 111 ms, 10 = 100 ms. A single number to hold in your head is easier than a rate.
4. Spread. Under about ±25 ms at a moderate speed is a held rhythm. Over ±60 ms means the run had visible stalls. Compare spread across runs of the same length only; it is not comparable across windows.
5. Dropped presses. Should be zero. If it is not, suspect double-firing hardware or software, not your finger.

The page also keeps your best per length and input, your last eight runs and a running click total in this browser. Bests are compared only within the same mode: a 1-second spacebar best never competes with a 60-second left-click best, because they are different tests. "Reset bests" clears the bests, the run history and the totals on this device; it does not clear the selected mode or the sound setting.

9. A practice routine that uses the numbers

The page is a measurement, so practise like you are using an instrument: one variable at a time.

1. Pick 10 seconds and left click. This is the default for a good reason: long enough for pacing, short enough to repeat without fatigue.
2. Set a click target, not a CPS target. 50 clicks in 10 seconds is 5.0 CPS. Count in blocks of ten: ten clicks, then ten more. Four blocks of ten plus ten is 50. Block counting keeps the rhythm while your hand does the work.
3. Run three times and look at the spread, not the best. If the three runs are within a few tenths of a CPS, you have a repeatable pace. If one is 5.6 and the next is 4.4, you are still burst-and-stall; slow down slightly until the three agree.
4. Then try 5 seconds. The same rhythm for half the time should give a similar CPS with a higher burst. If the 5-second CPS is much higher, your 10-second problem is endurance. If it is the same, your problem is top speed.
5. Only then try 60 seconds. Use it to see how much of your 10-second pace survives. The burst staying high while the average falls is normal; what you are training is the size of that gap.
6. Compare inputs last. Two runs per input, same length. Keep the input with the lower spread, then work on its pace.

Do not chase a single best. A personal best is one sample; the useful signal is the median of five runs at the same mode. That is also why the page keeps the last eight runs: the history is the measurement, the single card is just the latest sample.

10. FAQ

How many clicks per second is fast?

The page's own labels put 3 CPS at Brisk, 5 at Fast, 7 at Very fast and 10 at Blazing, so 5 CPS (50 clicks in 10 seconds) is the first "Fast" score and 7 CPS (70 clicks) is the next step. These are editorial labels for your own number, not a population percentile: the page has measured no other players and deliberately publishes no "average person" figure. Use them as rungs on a ladder, not as a claim about where you stand among other people.

Does the one-second test count the first click?

Yes — and that is the whole point. The first press starts the clock, counts as click one, and is the timestamp every later gap is measured from. There is no warm-up press. A 1-second run that reports 8 clicks is 8.0 CPS; the 8 includes the press that started it.

Why does the page drop presses, and what does "too fast" mean?

A press less than 12 ms after the last counted press is dropped and reported. A physical button cannot close twice inside 12 ms (USB polling is 1–8 ms), so the second event is a double-fire or a synthetic one. The page keeps the interval floor honest instead of silently counting it. If your own run shows dropped presses, suspect a failing mouse switch or a macro, not your finger.

Is right click faster than left click?

The page does not measure hardware, so it cannot answer that in general, and it does not pretend to. What you can do is run the same 10-second test twice on each input and compare the spread. The right button is usually a different switch with a different feel, and right-click mode is mouse-only because a touchscreen cannot send a secondary button. Pick the input you are more consistent on.

Can an autoclicker cheat this test?

A naive one cannot post an impossible number: presses are filtered at 12 ms and the drops are reported, so the ceiling for a synthetic stream is about 83–84 CPS. But the gate is an interval floor, not a software detector — a clicker paced at 5 ms still gets about 67 CPS counted. The page is a personal measurement, not an anti-cheat leaderboard; there is no leaderboard to cheat.

Why do my 60-second and 1-second scores disagree?

They measure different things. In a 1-second run every click is worth 1.0 CPS, so the start dominates and there is no time to recover. In a 60-second run every click is worth about 0.017 CPS, so the average is shaped by your slowest stretches and the burst stays high while the average falls. A gap between the two is normal; it is a fatigue curve, not a contradiction.

What does "spread" mean, and what is a good value?

Spread is the standard deviation of the gaps between counted presses, in milliseconds. Low spread at a given speed means you held a rhythm; high spread means you burst and stalled. There is no universal "good" number because it depends on your pace and the window, so compare only runs of the same mode: a spread under about ±25 ms at a moderate pace is a held rhythm, and over ±60 ms means visible stalls.

Does the page save my scores?

Your best per length and input, your last eight runs and your running click total are kept in this browser only, per mode. Nothing is uploaded and there is no account. "Reset bests" clears them on this device. A different browser, a private window or cleared site data starts from zero.

Why does the page not show a percentile or an average player?

Because it has measured no players. A percentile would require a population sample the page does not have, and inventing one would make the number look more authoritative than it is. The pace band is an editorial label for your own result, and the only comparison the page can honestly make is your own best on the same mode.