Accuracy before speed: why raw WPM flatters you

Net words per minute is raw speed multiplied by accuracy, so typing at 60 raw with 92 per cent accuracy gives 55 net — and the eight per cent you got wrong cost five words a minute before anyone corrected anything. In real work the cost is higher, because a mistake has to be noticed, found and fixed, and that takes longer than typing the character correctly the first time.

This is why practising for speed usually stops working. Pushing the raw figure up while accuracy falls leaves the net figure flat, and the typing feels worse.

What does the arithmetic look like?

The same raw speed at different accuracies:

Two of those rows are worth comparing. Someone typing 60 raw at 92 per cent and someone typing 57 raw at 100 per cent produce almost the same net figure — and the second one produces text that does not need fixing.

What is a word, in words per minute?

Five characters including the space that follows. It is a convention from the early 1900s, when typing speed was measured for hiring and the average English word plus its trailing space came to about five keystrokes.

The fixed definition is what makes the measure comparable at all. It also means a test on German text gives a lower figure than the same typist on English, because German words are longer and the yardstick is not — a real effect that has nothing to do with the typist.

Why does raw speed keep rising while net does not?

Because raw speed responds quickly to effort and accuracy does not. Pushing harder produces more keystrokes immediately, and the extra keystrokes include more errors, so the two effects cancel.

The way out is to make accuracy the target and let speed follow. Typing deliberately at a rate where errors nearly vanish trains the movements correctly; the speed then rises on its own as the movements become automatic. Practising at a rate that produces errors trains the errors.

A workable rule: if accuracy is below 97 per cent, slow down until it is not. The net figure will usually rise on the same day.

There is a second reason net is the honest figure. A test that penalises errors only by excluding them understates the cost, because in real work a typo has to be located as well as retyped — and locating one in a paragraph you have already moved past takes far longer than the keystroke did.

What is a good speed?

The average adult types around 40 words per minute. A proficient touch typist sits between 65 and 80, and professional transcriptionists exceed 100.

Context matters more than the number. Forty words a minute is entirely adequate for most jobs, because most typing is interrupted by thinking. The speed that matters for a developer or a writer is rarely the peak — it is whether the keyboard is fast enough to keep up with the sentence being composed, and that threshold is much lower than the numbers people compete over.

Keyboard layout is worth mentioning and worth deprioritising. Alternative layouts do reduce finger travel, and the gain is small compared with the cost of being unable to type on anyone else’s machine. Fix technique on the layout you have before considering a different one.

What actually improves it?

Four things, in the order they pay off.

1. Stop looking at the keyboard. Everything else depends on this, and it costs a genuinely unpleasant week.
2. Fix the home row habit. Consistent finger assignment is what makes speed possible; improvised fingering plateaus early.
3. Practise the awkward pairs. Most errors concentrate in a handful of letter combinations, and drilling those beats drilling random text.
4. Type real material. Prose you would actually write transfers better than word lists, because it contains the punctuation and capitalisation that break rhythm.

Short sessions beat long ones. Fifteen minutes daily produces more than two hours on a Sunday, because the skill is motor learning and motor learning consolidates between sessions rather than during them.

Questions people ask

Why is my raw speed higher than my net? Because raw counts everything you typed and net counts only what was correct. The gap between them is your error rate expressed in words.

Where do the words come from? A list of common English words, so the test measures typing rather than reading comprehension. Unusual words would test something else.

Is my best score stored anywhere? It stays in your browser. Nothing about what you type is sent anywhere.

Can I use a phone? You can, and the figures are not comparable to a keyboard. Thumb typing has its own ceiling and its own measure, and comparing the two says nothing useful.

Does the keyboard itself matter? Less than technique and more than nothing. Key travel and actuation force change how it feels and how fatigued you get over an hour, and neither will move a plateau caused by looking at your hands.

Chase the accuracy figure and the speed figure follows; chase the speed figure and neither moves. The typing speed test reports raw and net separately so the gap is visible, the word counter measures what you produced afterwards, and the time zone converter is for the call you were warming up for.