Methodology
Typing tests: how to test your typing speed properly
Different tests measure different things, and the same typist can score very differently across them for reasons that have nothing to do with ability.
Choose your test
Your typing speed is not one number
Typing speed is not a stable personal attribute. It moves with how long you have been typing, the time of day, and even the day of the week.
Across a long session. In a 120-minute study of 24 young (18–30) and 24 middle-aged (50–67) participants, speed decreased with time on task in both groups, visible as longer inter-key intervals. Accuracy declined in the young group but not the middle-aged one.2
In real workplaces — the interesting one. Both inter-key interval and backspace percentage rose with time on task. But in the morning workers first held their speed and paid for it in errors to correct; later they readjusted, and both declined. The effect on speed was more pronounced in the afternoon. Mondays and Fridays favoured speed, other days accuracy, with no cumulative decline across the week.3
Over a waking day. Across 366 participants and more than 45,000 typing sessions, speed peaked about 7.5 hours after waking — 0.098 SD above average — and fell to over 0.1 SD below by 15.3 hours awake.4
So one result is a snapshot taken under particular conditions. To track progress, test at a consistent time of day and compare like with like.
All of that research measures 30 minutes to two hours of continuous work, or hours awake.None of it establishes a measurable decline inside a one- to five-minute test. It is context for why your results move between days, not a prediction about what happens during a test.
alt: Two charts of typing performance over time on task: in the morning speed holds while errors rise; in the afternoon both speed and accuracy decline.
How our tests work
What we report, and why it is more than one number
Measuring accuracy properly needs three records, not two: the text you were given, the full keystroke stream including every backspace, andthe final text you produced.5
Most tests keep only the first and the last. Comparing final text against the target is blind to every mistake you made and then fixed.
So errors split in two — corrected and uncorrected, summing to a total error rate. They are different behaviours, and merging them loses information. Two typists finishing with identical clean text may have had very different runs, and only one of them is likely to repeat it.
alt: Three rows showing presented text, the full input stream including backspaces, and the final text, with corrected errors visible in the stream but absent from the result.
alt: A bar showing total error rate divided into corrected errors and uncorrected errors.
Why the test itself changes your score
Whether a test lets you correct determines which measurements are even possible.5 Ratatype pauses the character count on a typo until you fix it.6 Monkeytype offers Normal, Expert — which fails if you submit an incorrect word — and Master, which fails on a single incorrect keypress.7
Same typist, same ability, different protocols, different numbers.
Why the words you are given matter
Research phrase sets are built for high correlation with English letter frequency, so results stay comparable across studies. And letter pairs typed with different hands or fingers predict typing speed better than letter repetitions.1
So the mix of letter transitions materially moves the score. Two tests on different word sets are not quite measuring the same thing.
alt: Keyboard diagram contrasting a letter pair typed with alternating hands against a pair repeated on the same finger.
When a result is not a measurement
The 168,000-participant study excluded anyone with an error rate above 25%, along with those who did not finish — about 12% of participants in total.1 Below a certain accuracy, a speed figure stops describing typing ability at all.
What kind of typing these tests measure
These tests measure copy-typing from a screen — the research standard, but not the only way people type.
- Copy-typing — text stays visible. Low memory demand, but attention splits between source and output.
- From memory — intervals shorter than copying, error rate significantly higher.8
- Dictation — speed set by the audio pace, so it measures keeping up rather than capacity.9
- Composition — much longer intervals; writing engages processes copying does not.10
alt: Four typing tasks compared: copying, from memory, dictation and composition, each with relative speed and its limiting factor.
Sources
- Dhakal, V., Feit, A. M., Kristensson, P. O., & Oulasvirta, A. (2018). Observations on Typing from 136 Million Keystrokes. CHI ’18. doi:10.1145/3173574.3174220
- Time-on-task and typing performance in younger and middle-aged adults. Frontiers in Psychology (2018). doi:10.3389/fpsyg.2018.01113
- Wiechmann, D. et al. (2020). Typing performance and time on task in the workplace. PLOS One. doi:10.1371/journal.pone.0239984
- Smartphone typing speed across the waking day. PLOS Digital Health (2026). doi:10.1371/journal.pdig.0001281
- Soukoreff, R. W., & MacKenzie, I. S. (2003). Metrics for text entry research. CHI ’03. doi:10.1145/642611.642632
- Ratatype, on pausing the character count until a typo is corrected.
- Monkeytype documentation, on Normal, Expert and Master modes.
- Wallot, S., & Grabowski, J. (2013); Kristensson, P. O., & Vertanen, K. (2012).
- Abecassis, M., Magen, H., & Weintraub, N. (2026). Reading and Writing. doi:10.1007/s11145-025-10697-x
- Salthouse, T. A. (1986). Perceptual, cognitive, and motoric aspects of transcription typing. Psychological Bulletin.