Weak-Key Analysis: find the exact keys slowing your typing down
Most typing tests tell you a number. This one tells you which keys produced it. You type a passage built so that every letter of the alphabet appears several times, including the rare ones that ordinary English text barely uses, and the analysis reports the accuracy and response time of each key individually, which keys you confused with which, and then generates a practice drill aimed at the ones that cost you. Free, no sign-up.
About ninety words, chosen so that every letter of the alphabet turns up several times, including J, Q, X and Z. There is no clock. Type at your normal pace, because rushing produces errors caused by hurry rather than by the keys, which is the opposite of what this is looking for.
Click the text to begin. Nothing is timing you.
Where to go from here
Why the passage looks the way it does
Ordinary English is a terrible test of a keyboard. The letter E turns up about a hundred and twenty times more often than Z, so a normal passage will have measured your E thoroughly and your Z not at all. If a report on that basis stayed silent about Z, it would be leaving out the letter most likely to be your problem.
So the passage here is weighted: real words, chosen so that the rare letters each appear several times, shuffled in with common ones so it still reads as typing rather than as a spelling list. It is not a pangram, which only guarantees one appearance of each letter, and one appearance proves nothing.
What counts as a weak key
A key must be seen at least four times before this tool will say anything about it, and the count is printed next to every verdict so you can weigh it yourself. Below that threshold the arithmetic is not meaningful: one miss out of one appearance is a 0% accuracy that means nothing at all, and telling somebody their R is weak on that basis sends them off to practise the wrong thing for a fortnight.
Two things can mark a key as weak. The obvious one is errors: you pressed something else. The quieter one is hesitation, where you hit the key correctly but consistently took much longer to find it than you took for your other keys. The second is often the more useful finding, because it is invisible in an accuracy score and it is exactly what disappears when a key finally becomes automatic.
Keys the run did not see often enough are listed separately rather than dropped, so you can tell "this key was fine" apart from "this key was not tested".
Confusions, and why the pair matters more than the key
When the analysis can identify which key you pressed instead of the right one, it reports the pair. This is a genuinely different diagnosis. Missing R at random suggests you do not know where R is. Typing E where R was wanted, repeatedly, suggests your right index finger is anchoring one key to the left, which is a posture problem with a posture fix and nothing to do with knowing the alphabet.
The most common confusions are neighbours on the same finger, and mirror pairs across the two hands. Both are fixed by re-establishing the home-row anchor rather than by drilling the individual letter.
The drill this produces
The follow-up exercise is generated from your own weakest keys, using real words that contain them, with short repetition chunks between the words for the reps. You can run it immediately on the same page, as many times as you like.
This is a smaller version of what KeyTopia does continuously. Inside the app, the same per-key map is kept across every session you ever type, so the practice text is rebuilt from thousands of keystrokes instead of one passage, and it updates as keys move from weak to reliable. The tool here is the same idea with a much shorter memory.
Questions people ask
How many mistakes make a key "weak"?
It is a rate, not a count, and it needs at least four sightings of the key before any judgement is made. One miss out of one appearance is not evidence. The tool prints the attempt count next to every key so you can see how much weight the verdict deserves.
Why do I keep hitting the wrong key in the same place?
Usually because a finger has drifted off its home-row anchor, so every reach from it lands one key out. The confusion list on your result names the specific pair. Re-anchoring on the F and J bumps, and then re-typing the pair slowly, fixes it faster than drilling the letter on its own.
Can a key be weak if I never get it wrong?
Yes, and it is common. If you hit a key correctly but take twice as long to find it as your other keys, it is not automatic yet. That hesitation is invisible in an accuracy score and is often the single biggest thing holding a speed back.
How long does the analysis take?
The passage is about ninety words, so two to four minutes for most people. It is untimed on purpose: rushing it produces errors caused by hurry rather than by the keys, which is the opposite of what the analysis is trying to see.
Does this work on a Dvorak or Colemak layout?
The analysis works on any layout, because it reports on characters rather than on physical key positions. The confusion pairs are still meaningful, though the "neighbouring key" interpretation assumes QWERTY. KeyTopia proper is layout-aware and adjusts its lessons for QWERTY, QWERTZ, AZERTY, Dvorak and Colemak.
Further reading
A test measures. It does not teach.
KeyTopia turns the keys that slowed you down into the practice text for your next session, and keeps doing it as they improve. It is free, carries no advertising, and works without an account until you want your progress to follow you between devices.