What a "word" means here
The standard used across every typing test worth comparing against is five keystrokes to a word, spaces included. It does not matter what the actual words are: a passage of long technical terms and a passage of short common words produce comparable scores, because both are divided by five. Gross WPM is therefore total characters typed, divided by five, divided by elapsed minutes. CPM — characters per minute — is the same measurement without that division, and it is reported here as well because some keyboard and localisation tests quote it directly.
This convention is why WPM figures from a test and from a real writing session diverge. Composing prose involves pausing to think; a copy test does not. Someone who types 75 WPM here will produce far fewer than 75 words a minute of original writing, and that is not a fault in either number.
How accuracy is scored
Each character you type is compared with the character in the same position in the passage. Every mismatch counts as one error. If you stop early or overshoot the end, the difference in length is added to the error count as well, so leaving half the passage untyped is not a way to score highly. Accuracy is then one minus errors divided by the length of the passage, expressed as a percentage.
Position-based comparison has one consequence worth understanding: a single dropped character early on shifts everything after it by one place, and every subsequent character is then counted as wrong. That is deliberately harsh and it matches how the error actually behaves — a passage typed one character out of alignment is wrong for its whole length, and you would have had to fix it. If you notice a mistake, backspacing to correct it costs time but restores alignment, which is nearly always the better trade.
Net WPM, and why it is the number to quote
Net WPM here is gross WPM multiplied by accuracy. Ninety gross at eighty percent accuracy is seventy-two net; seventy gross at ninety-eight percent is sixty-nine. The two are nearly equal on this scale, and in real work the second is far more useful, because the first left eighteen errors somewhere in the text for you to find.
| Net WPM | Where that sits |
|---|---|
| Under 25 | Hunt and peck, or an unfamiliar keyboard layout |
| 25–40 | Comfortable non-touch typing, most casual computer users |
| 40–60 | Around the adult average for people who type daily |
| 60–80 | Trained touch typing, typical for professional office work |
| 80–100 | Fast; the upper end of what most people reach with practice |
| Over 100 | Uncommon, and usually the result of deliberate training |
What changes the number besides your hands
Passage content matters more than most people expect. Plain lowercase prose is the easiest case. Capitals cost a shift keystroke that does not count as a character. Punctuation, quotes and parentheses force finger movements away from the home row, and digits and symbols — the third passage type here — are where trained typists slow down most sharply, because number-row and symbol positions get far less practice than letters.
The keyboard itself is worth accounting for too. Switching between a laptop and a mechanical board, or between an ISO and an ANSI layout with a differently shaped Enter key, reliably costs several WPM for the first few minutes. So does a screen you are reading at an unusual distance. Run the same passage type on the same keyboard if you want to track your own progress, and ignore small differences between single runs — the variation between two consecutive attempts is easily five WPM in either direction.
Questions people ask
Why does the timer start late?
It starts on your first keystroke rather than when the passage appears, so reading time and getting your hands positioned do not count against you. It stops when you reach the end of the passage, or when you press Enter past the halfway mark, or immediately on Escape. Nothing runs before that first key, which also means you can sit and read the whole passage first if you want to.
Should I fix my mistakes or keep going?
Fix them. Errors are compared position by position, so a missed character shifts everything after it and every remaining character is scored wrong. Backspacing costs a second or two; leaving the misalignment costs the rest of the passage. This also matches real typing, where an error you leave behind has to be found and corrected later anyway.
Why is my score different from another typing site?
The five-keystrokes-per-word convention is nearly universal, but what happens after it is not. Some tests count only correctly typed characters toward gross speed, some subtract a fixed penalty per error instead of scaling by accuracy, some measure until the last correct character rather than the last keystroke, and passage difficulty varies enormously. Expect a spread of ten WPM or more between sites and compare each one only against itself.
Is anything about my typing recorded?
No. The passage, your keystrokes and the result exist only in this page while it is open. Nothing is sent to a server and nothing is stored between visits, which also means there is no history or leaderboard here — if you want to track progress, write the number down.