How WPM is calculated (and why tests disagree)
Words per minute sounds simple, but tests count it differently. Here is the standard definition and what can make two tests disagree.
A “word” is five characters
Typing tests don't count actual words, because words vary in length. The standard definition used in research treats every five characters, including spaces, as one word, and divides by the time in minutes (Dhakal et al., 2018). Typing 250 characters in one minute is 50 wpm.
Speed, raw speed and accuracy
- Speed (wpm) on Boost Type counts only characters you typed correctly. Mistakes don't add to it.
- Raw speed counts every keystroke, right or wrong, so it is always at least as high.
- Accuracy is the share of your keystrokes that were right the first time. A mistake you then fix still counts against accuracy, because it still cost you time.
Why tests give different numbers
- The text. Common letter pairs are typed faster than rare ones (Behmer & Crump, 2017), so a test of the 200 most common words gives a higher number than one with longer, rarer words or punctuation.
- The length. Short tests are noisier, and a few seconds of hesitation weigh more.
- How errors are counted. Some tests subtract errors, some only count correct words, some count raw keystrokes.
- Chance. Any single run varies by a few wpm.
Measuring real progress
To see whether you are really improving, use the same kind of text each time and compare your typical result over several runs. Boost Type's weekly benchmark uses text of the same difficulty every time, and only says you got faster when the change is bigger than your own normal variation.
Find your own weak spots
Boost Type records which keys, letter pairs and habits cost you the most time, then builds short practice around them. Three tests is enough to start.
Questions
How is WPM calculated?
Characters typed (including spaces) divided by five, divided by the time in minutes. 250 characters in one minute is 50 wpm.
What is raw WPM?
Speed counting every keystroke, including mistakes. Net or plain WPM counts only correct characters.
Why is my WPM higher on some sites?
Different texts, test lengths and ways of counting errors. Easy, common words give higher numbers.
Sources
- Dhakal, V., Feit, A. M., Kristensson, P. O., & Oulasvirta, A. (2018). Observations on typing from 136 million keystrokes. CHI '18. doi.org/10.1145/3173574.3174220
- Behmer, L. P., Jr., & Crump, M. J. C. (2017). Crunching big data with fingertips: How typists tune their performance toward the statistics of natural language. In Big Data in Cognitive Science (pp. 319-341). Routledge. www.crumplab.com/publications/Crump/files/4991/Behmer%20and%20Crump%20-%202017%20-%20Crunching%20big%20data%20with%20finger%20tips%20How%20typists%20t.pdf