WPM, accuracy, and consistency: how typing stats are actually measured

Typing tests hand you a pile of numbers — WPM, raw WPM, accuracy, sometimes a consistency score and a squiggly chart — and mostly don't explain any of them. That's a shame, because the math is simple, and once you know it, the numbers stop being a scoreboard and start being a diagnosis. Here's what each stat means, how it's computed, and what it's actually telling you to do.

Why is a "word" five characters?

The foundation of every typing stat is a standardization trick that dates back to typewriter-era typing contests: one "word" equals five characters, spaces included. Without it, scores would depend on which words the test happened to serve — a run of "it is so we do" would score double a run of "particularly necessary throughout." With it, every keystroke counts the same, and scores are comparable across tests, apps, and decades. So when a test says 60 WPM, it means you produced 300 correct-ish characters per minute — the test never counted dictionary words at all.

What is raw (gross) WPM?

Raw — or gross — WPM is pure output:

raw WPM = (all typed characters ÷ 5) ÷ minutes elapsed

Type 390 characters in 60 seconds and your raw WPM is 78, mistakes and all. Raw WPM measures finger velocity: how fast you can physically produce keystrokes. It's a real, useful number — it just isn't the one you should brag about, because it counts garbage output at full price.

What is net WPM?

Net WPM starts from gross and charges you for mistakes you left in the text:

net WPM = gross WPM − (uncorrected errors ÷ minutes elapsed)

Each error you didn't fix docks one "word" per minute from your score — a convention as old as typing certification, designed so that spraying errors can't inflate a score. Errors you did fix aren't penalized here; they cost you time instead (the backspacing shows up as a lower gross). The gap between your raw and net numbers is your error tax: raw 78 / net 71 means seven words a minute are being lost to mistakes, and accuracy work will pay better than speed work. Small print worth knowing: net WPM can't go below zero, and different apps count "uncorrected error" slightly differently (per character, per word), so compare net scores within one app, not across apps.

How is typing accuracy calculated?

Accuracy is the simplest stat: correct characters ÷ total characters typed. If you typed 400 characters and 388 matched the target, that's 97.0%. Two subtleties make the decimal matter. First, the scale is compressed — nearly everyone lands between 90% and 100% — so the difference between 96% and 99% is huge: it's the difference between one error every 25 characters and one every 100. A whole-number accuracy stat hides most of the signal, which is why serious tools report tenths. Second, accuracy pairs with net WPM to tell you which problem you have: high accuracy + low WPM means you can safely push speed; low accuracy + high raw WPM means you're practicing mistakes, and slowing down will make you faster within weeks. The training implications are laid out in how to type faster.

What does a consistency score measure?

Consistency answers a question the averages can't: was your pace steady, or a series of sprints and stalls? A test samples your speed repeatedly during the run (that's the data behind the pace chart), then measures how much those samples vary around their mean — statistically, a coefficient of variation, usually rescaled so that 100 means perfectly steady and lower numbers mean spikier. Two typists with identical 60 WPM averages can have wildly different consistency: 85 means a metronome; 55 means bursts of confidence separated by hesitation. Hesitation is the interesting part — those dips almost always mark patterns your fingers haven't automated: a rare letter combination, a capital, a number. Consistency is the stat most predictive of comfortable long-form typing, and it's trainable: deliberately even-tempo practice slightly below your maximum raises it within days.

How do you read the pace-over-time chart?

The pace chart plots your WPM (often raw and net as two lines) across the duration of the run. Three patterns to look for:

• Slow start, strong middle — normal; the first seconds include reaction time. Judge yourself on the curve after it settles.
• Sawtooth — repeated surge-and-stall. Cross-check with per-key stats or the words where dips happen; something specific is breaking your rhythm.
• Sagging tail — you fade on longer runs. That's endurance, and the fix is more 100-word or two-minute runs, not more sprints.

Why do personal bests need context?

One last measurement honesty rule: a WPM score only means something relative to its configuration. Fifteen-second sprints run hot (you can hold a burst that long), two-minute runs run cool, quotes score below common words because capitals and punctuation slow everyone down. A single global "best WPM" mixes all of that into a meaningless trophy. Track bests per mode, duration, and content — and compare today's 60-second words run against yesterday's 60-second words run, nothing else. Then check what is a good typing speed to see where your honest numbers land.