Actions per minute came out of real-time strategy, where the measurement had a job: estimate how much a player was doing at once. It counts every meaningful input — mouse clicks, key presses, hotkey combinations, box selections — and sums them over a minute.
Clicks per minute counts one button.
Why the numbers do not translate
A StarCraft player at 300 APM is not clicking 300 times a minute. They are issuing a mixture of clicks and keystrokes, and the keyboard usually contributes more than half of it. The same player’s honest CPM on a single button might be 350 — a number that sounds similar and describes something else entirely.
Going the other way is worse. A 900 CPM butterfly run is not 900 APM of play; it is one motion repeated, which no APM counter would ever see in a game.
| APM | CPM | |
|---|---|---|
| Inputs counted | Mouse + keyboard, all types | One mouse button |
| What it estimates | Workload across a match | Rate of one motion |
| Typical interval | A whole game | 1–60 seconds |
| Inflated by | Spam that changes nothing | Switch chatter |
The shared failure
Both metrics have the same weakness, which is that repetition inflates them without adding anything. APM has a name for it — spam clicking a rally point to pad the counter — and a partial fix, in “effective APM”, which discards actions that change no game state.
CPM has no equivalent fix, because in a click test the repetition is the measurement. What it has instead is the interval, which is the only honest way to compare two readings.
If you are training one
Train the one the game rewards. APM is a workload measure and improves with build-order fluency far more than with finger speed. CPM is a rate measure and improves with technique and hardware. Practising one will not move the other.