What WPM Actually Measures (and Why It's a Flawed Metric)
A score that had been climbing for weeks suddenly drops ten words a minute overnight, or two different typing tests measure the same fingers on the same keyboard back to back and land twenty words apart. Neither one means typing actually got worse. Words per minute sounds like it's counting words. It isn't. It's counting characters and dividing by a constant, and that one shortcut explains most of what the metric gets wrong.
A "word" is defined as five characters, not an actual word
Real words vary enormously in length, which would make WPM incomparable across different texts. A test full of short words like "a," "to," and "it" would produce an inflated score compared to a test full of long words, even at identical typing effort. Typing tests sidestep that by defining a "word" as exactly five characters, including spaces and punctuation, regardless of where actual word boundaries fall. WPM is then calculated as (total characters typed ÷ 5) ÷ minutes elapsed. A standardization convention that's been used since the mechanical-typewriter era specifically to make speeds comparable across different pieces of text.
That's also the simplest explanation for two tools disagreeing on the same typist: they're rarely scoring identical text. A passage loaded with short, common words divides differently against the five-character unit than one loaded with longer words, so the same fingers at the same speed can post different WPM numbers on different passages, with no actual change in skill behind either score.
Gross WPM vs net WPM
Raw typing speed by that formula is gross WPM, and it doesn't account for mistakes at all. A fast, sloppy typist and a fast, accurate typist can post the same gross score. Net WPM corrects for that: the common formula subtracts a penalty of one word (five characters) for every uncorrected error from the gross count before dividing by time, so a test riddled with errors scores substantially lower on net WPM even at identical raw keystroke speed. Most reputable typing tests report net WPM, or gross WPM alongside an explicit accuracy percentage, specifically because gross WPM alone is easy to inflate by typing carelessly fast.
This is usually where a sudden, alarming drop comes from too. A tired hand, an unfamiliar keyboard, or just a harder passage produces a few more uncorrected errors than usual, and net WPM penalizes each one directly. The raw keystroke speed underneath often barely moved; the score just stopped hiding the mistakes.
Transcription speed isn't composition speed
Nearly every typing test measures the same narrow skill: transcribing text you can already see, as fast as possible. That's a real skill, but it's a different one from the typing most people actually do most of the day. Composing original sentences while thinking through what to say, where the bottleneck is usually the thinking, not the keystrokes. A high WPM score on a transcription test says relatively little about how fast someone drafts an email or writes code, which is why WPM is a reasonable proxy for pure data-entry roles and a much weaker one for most knowledge work.
What to actually do with the number
If the goal is tracking whether you're actually improving, compare scores from the same kind of test over time rather than against a single outside benchmark; a consistent test is what makes the number meaningful, and net WPM specifically is the fairer measure since it already accounts for errors. If the goal is reconciling two different tools that gave two different numbers, stop treating either as the "real" score and check what each is measuring: gross or net, and how long and how varied the passage was. If the goal is judging real-world readiness for a data-entry role, net WPM on a reasonably long, varied passage is the closer proxy; a single short burst on easy text will read higher than sustained typing ever does.
Measure your own gross and net typing speed with the typing speed test.