Skip to the test
HumanBenchTests

Typing · net WPM and accuracy

Typing Test

Type the words below and this typing test reports your net words per minute — characters typed divided by five, minus your uncorrected errors, divided by the minutes elapsed. Accuracy, characters per minute and a per-key error heatmap come with it. Pick a length, start typing, and the clock starts on your first keystroke.

1 minute · common words

Time left

1:00

Net WPM

0.0

Accuracy

0.0%

Errors

0

0 correct

Click or tap here and start typing — the clock starts with your first keystroke

Ready — 1 minute. Type the words above; the clock starts when you do.

Per-key error heatmap

No mistyped characters yet

qwertyuiop
asdfghjkl
zxcvbnm
space

How is a typing test scored?

Every typing benchmark since the mechanical typewriter has defined one word as five characters, spaces included. That convention exists so a score does not depend on whether the passage happened to contain long words. From it follow the only two formulas you need:

gross WPM = (characters typed ÷ 5) ÷ minutes
net WPM   = (characters typed ÷ 5 − uncorrected errors) ÷ minutes
accuracy  = correct characters ÷ characters typed × 100

Gross WPM ignores mistakes completely, which is why a test that quietly reports it reads 5 to 15 points higher than one reporting net. Net WPM subtracts one whole word for every uncorrected wrong character — a deliberately harsh penalty, because in real work an error costs far more than the second it took to make.

A worked example. You type 312 characters in exactly one minute and leave 4 characters wrong. That is 312 ÷ 5 = 62.4 words gross, so gross WPM is 62.4. Subtract the 4 errors and net WPM is 58.4. Accuracy is 308 ÷ 312 = 98.7%. The result panel above does this with your own numbers rather than making you take a figure on trust.

What is a good score on a typing test?

Speed bands used across this site
Net WPMBandWhat it means
0–20Hunt and peckTyping by looking at the keys. Almost all of the gain here comes from touch typing.
20–35Below averageFunctional for messaging, slow for long-form writing or note taking in meetings.
35–45AverageThe band an average adult lands in. Around 40 WPM is the number usually quoted.
45–60Above averageComfortable for office work. You are typing faster than most people you work with.
60–80FastMeets the 60 WPM floor most data-entry and administrative job adverts specify.
80–100ProfessionalTranscription territory. Sustained accuracy matters more than peak speed up here.
100+EliteTop ~1% of typists. Rare to sustain over a five-minute test without accuracy loss.

For where a specific number sits in the wider population, the average typing speed by percentile page has the full distribution table. If you only want a quick figure without picking a mode, use the 25-word speed check.

How to take this test properly

  1. 01

    Pick a length

    One minute is the comparable standard. Word modes stop when you finish the passage; timed modes stop when the clock does.

  2. 02

    Start typing

    The clock starts on your first keystroke, not on a countdown, so there is nothing to react to and no wasted tenths.

  3. 03

    Correct or carry on

    Backspace works. A correction costs you time but not accuracy; leaving the error costs you a word off your net WPM.

  4. 04

    Read the arithmetic

    The result panel expands the formula with your own character count, error count and elapsed time so you can check the number yourself.

Three things distort a typing score more than anything else: the keyboard (a low-travel laptop board typically costs 5–10 WPM against a mechanical one), the chair and wrist position, and whether you are reading ahead. Good typists are usually two to three words ahead of their fingers; if you find yourself reading one character at a time, that — not finger speed — is the ceiling.

Typing test questions

How is the typing test score calculated?

Net WPM = (characters typed ÷ 5 − uncorrected errors) ÷ minutes elapsed. Five characters counts as one word, spaces included. Accuracy is correct characters ÷ characters typed. The test shows the whole calculation with your own numbers substituted in once you finish, so nothing about the score is a black box.

How long should a typing test be?

One minute is the standard for a comparable figure and it is what most employers use. Sprints of 10 or 25 words measure burst speed and flatter almost everyone. Five minutes measures something different again — sustained accuracy under fatigue — and typically comes out 5 to 10 WPM below a one-minute score for the same typist.

Do mistakes I correct count against me?

No. Errors are counted on the final text, so a character you backspaced and fixed is not penalized — the time it cost you already shows up in your WPM. That is how typing certificates are scored. Uncorrected mistakes, the ones still wrong when the clock stops, come off your net WPM at one word each.

What is a good typing speed?

Around 40 WPM is the figure usually quoted for an average adult. Above 60 WPM meets the floor most data-entry and administrative job adverts specify, and above 80 WPM is professional transcription territory. Below 30 WPM the fastest route to improvement is learning to touch type, not practicing faster.

Why is my WPM lower here than on other typing tests?

Two reasons, usually. Some tests report gross WPM and call it WPM, which ignores errors entirely and reads 5 to 15 higher. And prompt difficulty varies enormously — a test drawn from the 200 most frequent English words produces a higher number than one with punctuation, capitals and rare words. This test reports net WPM on a common-word prompt and says so.

Does the per-key heatmap actually help?

It is the most actionable output here. The heatmap is keyed by the character you were supposed to hit, not the one you actually pressed, so it points at a finger placement rather than a symptom. Most people find their errors cluster on two or three keys — very often the same-finger reaches like C, X and the number row.

Other instruments on this bench

Testing the hardware rather than the person? device tests — scroll, touch, refresh rate and input latency lives on a sister site. And if you want to see through someone else’s eyes rather than screen your own, use the simulate how an image looks to a protan, deutan or tritan viewer.