← Back to Blog

If you have ever taken a typing test, you have seen two numbers: WPM and CPM. They look like they measure the same thing in different units, and in a sense they do — but the relationship between them is more interesting than it first appears, and the convention behind WPM hides a tidy little piece of arithmetic history that almost nobody explains.

This is the short, honest explainer for what these numbers actually mean, how they relate to each other, and why the entire framework starts to look strange once you ask what it is actually measuring.

The Definitions, Plainly

CPM stands for characters per minute. It is the literal count of how many characters you produce in sixty seconds — every letter, every space, every punctuation mark. If you type the sentence "the cat sat on the mat" in twelve seconds, you have produced 22 characters in 12 seconds, which is 110 CPM. The math is obvious, and CPM is exactly what its name says.

WPM stands for words per minute. You would expect it to be a count of actual words. It is not. WPM is calculated by taking your CPM and dividing it by five. That is the convention. Every typing test, every speed leaderboard, every employer benchmark — when they report WPM, they mean (characters per minute) divided by five.

So the relationship is simple arithmetic:

If you score 60 WPM on a test, you typed 300 characters in that minute. If you score 100 WPM, you typed 500 characters in that minute. There is no actual counting of words involved.

Why Five?

The "five characters equals one word" convention dates back to the early days of typewriter speed testing in the late nineteenth and early twentieth centuries. The reasoning is straightforward: average English words are roughly five characters long, plus a space, which makes a five-character unit a reasonable proxy for "a word in standard prose."

Doing the count this way had two enormous practical advantages back when typing tests were administered by hand. First, it removed the need to argue about edge cases — is "doesn't" one word or two? does a hyphenated compound count as one or two? does a number written out as digits count? With a character-based count, there is nothing to argue about. You count characters, you divide by five, you have a number.

Second, the character-based count made tests comparable across different texts. If a test uses long words, the WPM count comes out lower for the same number of keystrokes. If it uses short words, the WPM comes out higher. By using a fixed character-to-word ratio, the convention controls for this variation and reports something closer to "raw keystroke speed measured in word-equivalents."

The convention is so universal at this point that nobody bothers to explain it. If you see a typing test reporting "65 WPM," that almost certainly means 325 characters per minute, regardless of whether the test actually contained 65 dictionary words or 80 or 50.

The Hidden Distortion

Once you know about the five-character convention, a small distortion becomes visible. Different texts produce different apparent WPM scores for the same actual typing speed.

Consider two passages of the same length in characters. The first is full of common short words: "the cat sat on the mat and ate a bit of fish." The second uses longer, more uncommon words: "philosophical considerations occasionally precipitate substantial revisions." Both passages are roughly the same number of characters. If you type each one in exactly thirty seconds, the WPM score will be roughly the same.

But the first passage contains eleven actual dictionary words. The second contains six. The "real" WPM, if you count actual words, is dramatically different — and yet the test reports the same number, because the convention does not care about real words.

This is fine for measuring keystroke speed, which is what typing tests are actually trying to measure. It is misleading if you think you are measuring "speed at producing meaningful text." Most typing tests use carefully curated common-word lists precisely because those lists produce flattering WPM scores, with average word lengths close to or below five characters.

Net WPM vs Gross WPM

There is one more wrinkle worth knowing. Many typing tests distinguish between gross WPM and net WPM.

Gross WPM is the raw calculation: total characters typed divided by five, divided by minutes elapsed. It does not account for errors.

Net WPM penalizes you for mistakes. The most common formula is:

Net WPM = Gross WPM − (errors / minutes elapsed)

So if you type at 80 gross WPM but make ten errors in a minute, your net WPM is 70. The point of net WPM is to discourage sloppy fast typing — there is no real benefit to producing 100 typo-ridden words a minute if the typos all have to be corrected later.

Different tests use slightly different error penalties, which is one of the reasons that the same person can get noticeably different scores on MonkeyType, 10FastFingers, and TypeRacer. They are not all running exactly the same formula. If you are comparing your scores across platforms, pay attention to which metric each one uses.

What KPH Means, And When You Care

You will occasionally encounter a third metric: KPH, or keystrokes per hour. KPH is the standard measurement used in data entry and clerical work, where the relevant skill is not narrative typing but rapid, accurate entry of numbers, codes, and tabular data.

The conversion is straightforward arithmetic:

So a typist clocking 60 WPM in narrative text is producing approximately 18,000 KPH at that pace. Data entry job postings commonly require 8,000-12,000 KPH for basic positions and 15,000+ KPH for senior data entry roles, which corresponds to roughly 27-50 WPM.

KPH is the better metric for data entry because the work involves many short fields — addresses, dollar amounts, dates, codes — where the "word" abstraction does not really apply. You are entering keystrokes, and the question is how many keystrokes per hour you can sustainably produce with acceptable accuracy.

A Worked Example

Let us run through an actual conversion to make this concrete. Suppose you take a one-minute typing test and the results show:

Your gross WPM is 425 / 5 = 85 WPM. Your CPM is 425. Your KPH at that pace is 425 × 60 = 25,500. Your net WPM (using the standard formula) is 85 − 3 = 82 WPM. All four numbers describe the same minute of typing, in different units, with slightly different things accounted for.

You can see why people get confused by typing metrics. The same performance can look quite different depending on which number is reported, and the conventions are not always made explicit.

What the Numbers Are Really Measuring

Step back from the arithmetic for a moment. What are these metrics actually trying to capture? They are all proxies for the same underlying question: how fast can your fingers move characters from your brain to a screen?

WPM, CPM, and KPH are all keystroke-velocity measurements wearing different units. They are useful because keystroke velocity is the binding constraint on typing speed for most people. They are also somewhat misleading, because they treat "characters produced" as if it is equivalent to "useful text produced," which is only true if your characters are correct, well-edited, and not duplicated by autocorrect.

But the bigger issue with all of these metrics is that they are framed around a specific input device — the keyboard — that has hard physical limits. The numbers exist because we have spent a century optimizing the act of moving fingers against a flat surface of keys. The fastest typists in the world top out around 200 WPM. The average office worker sits around 40 WPM. The realistic practiced-typist range is 60-100 WPM.

The Metric That Breaks the Frame

None of these metrics apply meaningfully to voice dictation. WPM, CPM, KPH, net WPM — they are all calibrated to the physics of finger motion against keys. Voice operates in an entirely different physical regime, and the resulting throughput is correspondingly different.

The relevant reference point is conversational speech speed. The average English speaker produces 130 to 160 words per minute in normal conversation. Trained speakers — auctioneers, rapid-fire podcasters, public speakers — sustain 200+ WPM. These are not records or sprints. They are ordinary, sustainable rates.

Converted to the keyboard metrics that typing tests use, conversational speech runs at:

The voice numbers are not slightly higher than the keyboard numbers. They are categorically higher, in a way that is invisible if you only think in terms of typing-test scoreboards.

Voice Keyboard Pro and the Numbers

The reason this comparison matters is that modern voice dictation has finally caught up with the throughput that human speech makes available. The latest generation of AI speech recognition operates above 95% accuracy on natural conversational speech, handles ambient noise and accents gracefully, and returns text in under a second of latency.

Voice Keyboard Pro is a native macOS app that brings that speed into practical use. It lives in your menu bar at just 1.7MB. You hold a hotkey, you speak, you release — and text appears at your cursor in whatever application you happen to be using. The transcription runs through Voice Keyboard Pro's advanced speech recognition on fast cloud infrastructure, with an optional offline mode powered by Apple's on-device Speech framework when you need to work without a network connection.

Translated into the metrics this article has been about: a comfortable speaking pace through Voice Keyboard Pro produces text at roughly 150 WPM, which is 750 CPM, which is 45,000 KPH. Without training. Without practice. On the first use.

The WPM metric only matters if your fingers are doing the work. Once your voice is the input device, the entire keyboard scoreboard becomes irrelevant.

When the Numbers Still Matter

The keyboard metrics are not going away, and they should not. They still measure something real for tasks that require precise keyboard input — coding, spreadsheets, command-line work, careful editing, anything where individual keystrokes carry meaning beyond raw text. A 60 WPM coder is genuinely more productive than a 30 WPM coder, and the keyboard-based metric captures that difference correctly.

But for tasks where the underlying goal is just "produce a lot of text quickly" — emails, drafts, notes, messages, documents — the keyboard scoreboard is a slow path. Voice Keyboard Pro has a free tier. Open it, dictate a paragraph, count the characters when it is done. The numbers will land somewhere the keyboard cannot follow.

WPM, CPM, KPH — they are all measurements of the same physical bottleneck. Once you stop typing, the bottleneck moves, and the numbers stop describing the territory.

Keep reading