Here's the thing that surprises most people about typing tests: WPM doesn't count words. If it did, a test full of "a", "to", and "it" would score triple a test full of "keyboard", "sentence", and "understand", and no two results would be comparable. So the industry standardized the word away.

A "word" is 5 characters

By convention, one word = 5 characters, including spaces. Type 300 characters in a minute — any characters — and you typed 60 "words". This is why WPM comparisons work across different texts and even different tests: everyone is counting keystrokes in disguise.

The base formula:

Gross WPM = (all typed characters ÷ 5) ÷ minutes

Gross WPM measures pure output — how much your fingers produced, right or wrong.

This isn't a house convention

It's worth knowing the five-character word is the actual research standard, not something typing sites made up to flatter scores. The largest published study of typing behaviour, Observations on Typing from 136 Million Keystrokes (Aalto University / University of Cambridge, CHI 2018), computes WPM the same way when reporting its headline result — an average of 52 WPM across 168,000 participants.

The convention predates computers. It comes from the typewriter era, where a standardized "word" was the only way to make results from different passages comparable, and it survived because the alternative — counting real words — makes a test of short words score triple a test of long ones.

Net WPM: where errors get their bill

Raw output isn't the number anyone cares about, because typing garbage fast isn't typing. The classic correction:

Net WPM = Gross WPM − (uncorrected errors ÷ minutes)

Each error you leave in the text costs you one full "word" per minute of penalty. Note the word uncorrected: mistakes you backspace and fix don't reduce net WPM (they already cost you time, which is punishment enough). Only errors still present when the timer stops count.

Uncorrected errors come in three flavors, and TypeTimer reports all of them separately on your results screen:

Incorrect — you typed a character, it was the wrong one, and you left it.
Extra — you typed characters beyond the end of a word.
Missed — you jumped to the next word before finishing the current one; the skipped characters count against you.

A fully worked example

You take a 1 minute test and type 312 characters. When the timer ends, there are 6 wrong characters you never fixed.

Gross WPM = (312 ÷ 5) ÷ 1 = 62.4 → 62 gross WPM
Error penalty = 6 uncorrected errors ÷ 1 minute = 6
Net WPM = 62.4 − 6 = 56.4 → 56 net WPM

Same performance on a 2 minute test (624 characters, 12 errors): gross = (624 ÷ 5) ÷ 2 = 62.4; penalty = 12 ÷ 2 = 6; net = 56. The math scales cleanly with duration — which is exactly why it's the standard.

Notice how expensive errors are: six stray characters cost this typist six whole WPM. At the margins, fixing your last few errors is worth more than typing faster. This is the entire argument behind accuracy-first training, and it's why the typing accuracy test leads with accuracy instead of speed.

Accuracy is a different number

Accuracy answers a different question: of every key you pressed, how many were right at the moment you pressed them?

Accuracy = correct keystrokes ÷ total keystrokes

Corrected mistakes still count against accuracy — you did press the wrong key, even if you cleaned it up. So you can finish with a high net WPM (everything fixed) and mediocre accuracy (lots of fixing). That combination means your speed is real but expensive; smoothing out the errors would raise it further.

CPM, KPH, and other units you'll meet

Unit	What it counts	Conversion	Where you'll see it
Gross WPM	(characters ÷ 5) ÷ minutes	—	Every typing test
Net WPM	Gross − (uncorrected errors ÷ minutes)	—	Serious tests, certificates
CPM	Characters per minute	1 WPM = 5 CPM	Non-English tests, where word length varies
KPH	Keystrokes per hour	1 WPM ≈ 300 KPH	Data-entry job listings
Accuracy	Correct keys ÷ total keys pressed	—	Reported alongside WPM

Two footnotes on the conversions. CPM is just WPM without the ÷5, so 60 WPM = 300 CPM exactly. KPH is looser: 50 WPM ≈ 15,000 KPH of raw output, but data-entry KPH tests often measure numeric-keypad work, which is a different skill with a different ceiling — don't assume your prose WPM transfers.

Why your WPM changes between tests

Two tests can apply this exact math and still disagree, because the input differs: random common words type faster than punctuated prose; some tests forbid backspacing; some end a word the moment you hit space, others let you return. None of this is cheating — they're just different measurement conditions. We wrote a full breakdown of why your WPM differs on every typing test.

See the math run on your own typing

Every TypeTimer result shows the whole chain: gross WPM, net WPM, accuracy, CPM, and the character-level counts (correct / incorrect / extra / missed) that feed the formulas — so you can audit your own score against the formulas above. Take a 1 minute test and check the arithmetic yourself; it's oddly satisfying when it lines up.

How Is WPM Calculated? The Formula