What does "average typing speed" actually mean?
An average is only as meaningful as the three things behind it, and almost every figure you will find online omits all three.
- Average of whom? A hundred and sixty-eight thousand people who searched for a typing test are not the same population as a hundred and sixty-eight thousand people picked at random.
- Measured how? Net WPM and raw WPM can differ by ten points on the same run. A test that ignores uncorrected errors reports a higher number than one that does not.
- On what text? Common English words with no punctuation are far quicker than a literary quote with capitals, commas and apostrophes in it.
Change any one of those and the average moves. That is why "the average typing speed is X" is not a fact so much as a summary of a specific measurement — and why the version repeated without a study behind it should be treated as folklore rather than data.
What is the best-sourced figure available?
52 words per minute, from researchers at Aalto University and the University of Cambridge who collected 136 million keystrokes from 168,000 volunteers across more than 200 countries (Dhakal, Feit, Kristensson and Oulasvirta, "Observations on Typing from 136 Million Keystrokes", published in the Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, DOI 10.1145/3173574.3174220). The fastest participants reached around 120 WPM.
It is the largest published typing dataset there is, and its caveat has to travel with it: participants were self-selected online volunteers, not a representative sample of any population. They opted into a typing test. That group skews toward people who type a lot and care how fast they do it, which almost certainly pushes the average above whatever the true population figure is.
Two other findings from the same dataset are worth knowing, because they change how you should read any average. Participants who had previously taken a formal typing course had similar typing behaviour to those who never had — similar speed, similar hand use, similar errors. And what distinguished the fastest typists was not per-key speed but overlap: they performed between 40 and 70 percent of keystrokes with the next key going down before the previous one came up. An "average" collapses all of that into one number and throws away the part that explains it.
Why do quoted averages vary so much?
Because the tests are not measuring the same thing. Five variables move a typing score before your fingers are involved at all.
| Variable | What it does to the number |
|---|---|
| Who took the test | The largest effect by far. Volunteers on a typing site are self-selected; office workers pulled into a study are not. |
| Net or raw WPM | Raw counts every keystroke; net counts only characters correct in the final text. Raw is always the higher of the two once you make a mistake. |
| How errors are charged | Some tests subtract uncorrected errors, some ignore them, some block you from advancing until the word is right. Each rule produces a different score from identical typing. |
| Text difficulty | The thousand most common English words with punctuation and numbers switched off are the easiest thing you can be asked to type. Quotes, symbols and digits all cost speed. |
| Test length | A 15-second run rewards a lucky burst; a 120-second run exposes whether the pace holds. Averages built from short tests run higher. |
There is a sixth difference that is invisible unless you go looking for it: research practice and product practice do not use the same clock. In text-entry research the character count is decremented by one before computing speed, because timing starts at the first keystroke rather than before it (I. Scott MacKenzie, "Measuring Text Entry Speed"). Typing-test products, Keyboardo included, divide the full character count by five over wall-clock elapsed time. On a 300-character minute that is the difference between 59.8 and 60.0 WPM — small, but it is one more reason two honest tests disagree.
How do you compare two typing speeds fairly?
Hold the measurement constant and only then compare the people. In practice that means matching five things: the metric (net or raw), the duration, the text source, the punctuation and numbers settings, and the language.
This is not a theoretical concern, and you can see it enforced in the way any serious leaderboard is scoped. Keyboardo's leaderboard only accepts results from English 1k word lists with punctuation and numbers switched off, taken live rather than imported, and not flagged by the anti-cheat checks. Everything else — quotes, French, Arabic, punctuation on, numbers on — is still recorded on your own profile but is deliberately kept out of the ranking, because a French quote run with punctuation is not the same test as an English 1k run without it and ranking them together would produce a table that means nothing.
Apply the same discipline to yourself. If you want to know whether you are improving, compare 60-second English runs to 60-second English runs. Comparing today's 15-second sprint to last month's two-minute run tells you about the format, not about you.
What about older figures, and speed by age?
Historical numbers are useful as long as they are dated. Undated numbers are how a reasonable 1999 measurement becomes a confident 2026 claim.
- 1999: average computer users measured at 32.5 WPM transcribing and 19.0 WPM composing, per the figures cited in Wikipedia's summary of typing-speed conventions. The transcribing-versus-composing gap is the durable part; the absolute values are a quarter of a century old.
- 1970s and 1980s: professional typists commonly 60 to 90 WPM, as noted by the Aalto and Cambridge researchers above.
- 2019: average mobile-phone typing measured at 36.2 WPM, also from the Wikipedia summary. Different input method, different skill, not comparable to a physical keyboard.
On age, the answer is shorter: no reliable public dataset breaks typing speed down by age. The tables that circulate — a figure for ages 6 to 11, another for 12 to 16, another for adults — have no traceable origin. They are copied from site to site until the repetition itself starts to look like evidence. They are the highest-risk item on this whole topic precisely because they are so tidy, and Keyboardo will not publish one until there is something real to publish.
What could a typing site honestly publish?
Its own data, described accurately. Keyboardo stores completed results, so a distribution computed from them — a median, deciles, the number of results, the date range — would be a real measurement rather than a repeated rumour, and it would be reproducible because the formulas are published.
It would still not be an average typing speed. It would be the average speed of people who came to a typing-test site and finished a test, which is the same self-selection that qualifies the 52 WPM figure, plus a second layer: practice sites attract people who practise. Any site publishing its own distribution and calling it "the average typing speed" would be making the identical mistake in a more sophisticated way.
Until such a page exists here, the honest position is the one at the top of this article. There is one credible published figure, it comes with a population caveat, and everything else in circulation is either dated, undated, or invented. If you want to know where you personally sit, the only reliable answer is your own number from a timed test taken the same way twice — and what counts as a good typing speed covers how to read it once you have it.