Measure your typing speed against the standard the whole field uses: a word is five characters including spaces, not an actual word. That convention exists because real words vary in length, so counting words makes a passage of short ones look far faster than the same keystrokes spread over long ones — and produces a figure you cannot compare with any other test. You get gross speed, net speed with uncorrected errors deducted, and accuracy scored against what you actually typed rather than against the length of the passage. Four passage types, because typing prose and typing code are different skills. Nothing is transmitted and no score is stored.
| Net WPM | Level |
|---|---|
| Under 25 | Beginner |
| 25–40 | Below the ~40 average |
| 40–60 | Average to good |
| 60–80 | Fast — professional typist |
| 80–100 | Very fast |
| 100+ | Exceptional |
Five characters including spaces. That is the standard everywhere, and it exists because actual words vary so much in length that counting them makes the number meaningless across passages. A test counting whitespace-separated words gives you a figure that compares with nothing.
Gross is everything you typed; net deducts uncorrected errors. Net is the one worth quoting — typing quickly and wrongly is not typing quickly, and most published benchmarks are net figures.
The adult average is around 40 words a minute. Sixty is solidly good, eighty is professional typist territory, and past a hundred is exceptional. Accuracy matters more than raw speed for most real work, since fixing mistakes costs more time than typing slightly slower would have.
Correct characters as a share of what you typed, not of the passage length. Stopping halfway through with everything correct gives you 100% accuracy rather than 50% — you typed correctly, you just typed less.
It fixes the error, so accuracy and net speed improve — but it costs you time, which lowers gross speed. That trade is real, and it is why the two numbers are shown separately.
Because they are genuinely different skills. Someone fast on prose is often much slower on symbols and digits, where the hands leave the home row constantly. Testing only prose hides that.
No. Nothing is transmitted and nothing survives a reload.
A word is five characters including spaces, and it matters that this is the standard rather than a shortcut. Real words vary enormously in length, so counting whitespace-separated words makes a passage of short words look far faster than the identical keystrokes spread over long ones — a test reporting words-per-minute by counting words produces a number that cannot be compared with any other typing test, or with the same test on a different passage. Five characters per word is the convention the whole field uses precisely because it is insensitive to that. Two figures are reported. Gross speed is every character typed divided by five, over the elapsed minutes. Net speed deducts uncorrected errors, which is what most tests headline, because typing quickly and wrongly is not typing quickly — and it is floored at zero, since a burst of mistakes can exceed the gross figure and a negative typing speed is not a thing. Accuracy is scored against what was typed rather than against the length of the passage, so stopping halfway does not report fifty per cent accuracy on work that was entirely correct. A character typed past the end of the passage counts as wrong rather than being ignored. The elapsed time is measured from the first keystroke rather than from when the page loaded, and from a monotonic clock rather than the system one, so a clock adjustment cannot produce a negative interval mid-test. Per-character state drives the colouring, which is computed from the two strings rather than tracked incrementally — tracking it incrementally means backspacing has to unwind state correctly, and getting that subtly wrong leaves stale colours that make the test feel broken. Nothing is transmitted and no score is stored anywhere.