The shape of a session
A practice session that produces improvement has to do two incompatible things: let you type badly on purpose, and produce one number you can trust. Trying to do both at once is why most typing practice turns into an hour of taking the same test and hoping. Separating them is the whole idea behind the four-part loop below.
| Part | What to run | What you are watching |
|---|---|---|
| Warm-up | Zen mode, or a 10-word run | Nothing. Do not score it. |
| Focused block | 25-word runs, or 30-second timed runs, with one constraint switched on | Accuracy, and nothing else |
| Measured test | One 60-second English run, punctuation and numbers off | Net WPM and consistency |
| Review | The results screen you just produced | Which words broke, and when in the test |
The order matters more than the durations. Everything before the measured test is allowed to be messy; the measured test is not allowed to change.
The warm-up
The first minute of typing is reliably your worst, and there is no reason to record it. Zen mode is built for this: it removes both the timer and the target text, so there is nothing to fall behind and nothing to get wrong. Because there is no target, net WPM equals raw WPM and accuracy is fixed at 100 — a zen run is deliberately not a measurement, and it never counts toward a personal best, a leaderboard or an achievement.
Use it to check the mechanics rather than the output. Are your hands anchored on the home row, are your wrists straight and in line with your forearms, is the keyboard in front of you rather than off to one side? Two minutes of typing anything at all, badly, with correct posture, is a better warm-up than a fast 15-second test that starts the session by telling you a number.
The focused block
This is the part of the session that actually changes anything, and it works on one principle: pick a single constraint, hold accuracy, and let speed be whatever it is.
A constraint is a deliberate handicap that forces the sequences you have not automated to appear more often. Keyboardo gives you several, and they stack:
- Punctuation on. Commas, apostrophes and capitals — every one of them a reach away from home row, and every one of them a Shift key you may be pressing with the wrong hand.
- Numbers on. The number row is the reach most typists never learned, and it is the one that costs the most in real work.
- Quote mode. Real sentences with natural punctuation and capitalisation, in short, medium or long lengths, instead of a shuffled word list.
- A different language. English 5k widens the vocabulary beyond the thousand most common words; French and Arabic — the latter with full right-to-left support — break English letter-pattern muscle memory entirely.
Turning on punctuation or numbers makes a result ineligible for the leaderboards, which require English 1k with both toggles off. Treat that as the feature it is: the toggles are how you mark a run as practice rather than competition, and nothing you do in this block will ever contaminate the boards.
The rule for the block is that accuracy governs pace, not the other way round. If you cannot hold the accuracy you would accept in real writing, the constraint is too hard or the pace is too high; drop one of them rather than accepting the errors, because errors typed repeatedly are practice too.
The one measured test
End every session with exactly one scored run, under a condition you never change. A 60-second English test with punctuation and numbers off is the sensible default, for two reasons: it is long enough that a single lucky burst cannot carry it, and it is the exact condition the all-time and daily leaderboards use, so the same run that measures you is also the run that could rank you.
There is a second reason not to pick something shorter. Keyboardo's speed, accuracy and consistency achievements only count on a test of a reasonable size: a live, unflagged run of at least 30 seconds in time mode, at least 25 words in word mode, or a medium or long quote. Zen runs never qualify, and imported results never qualify. A 15-second test is a fine warm-up and a poor measurement, and the site treats it that way.
Whatever you pick, write it down and keep it. A progress series assembled from 30-second runs, 60-second runs and quote runs is not a progress series; it is four series interleaved.
The review
Sixty seconds of reading the results screen is worth more than another test. Three things on it are diagnostic, and they point at different problems.
The gap between raw and net WPM is your error cost in units you can feel. Multiply it by five to get characters. Score raw 80 and net 68 and the 12 WPM gap is 60 characters a minute that did not survive into your final text — roughly a full line of prose per minute, thrown away. If that gap is wide, nothing about your finger speed is the problem.
Accuracy is keystroke-based and unforgiving: a character you typed wrong and then corrected still counts as an incorrect keystroke, because the correction repaired your text, not the record of what your hands did. It is the number that tells you whether the focused block is set at the right difficulty.
Consistency is built from your speed second by second, so it measures pauses rather than pace. A high WPM with low consistency means most of your output arrived in a few good seconds and the rest of the test was spent recovering.
Below the numbers, the screen lists the specific words you got wrong and marks your errors on the speed chart against the timeline. That is the material for tomorrow's focused block. A plateau usually has three or four named culprits in it, and they are on that list.
How often, and for how long
There is no researched optimum here, and this page is not going to invent one. No published study establishes an ideal typing-practice session length, an ideal frequency, or a rate of improvement per hour practised, and every confident figure you will find for those has been made up somewhere upstream.
What can be said honestly is structural. The loop above takes as little as five to ten minutes, which matters because the binding constraint on practice is almost never session length — it is whether the session happens at all on a Tuesday you did not plan for. A short routine you repeat beats a long one you abandon, and the practical test of any schedule is simply whether you can do it again tomorrow.
Use the daily challenge as the anchor
If you want the decision made for you, the daily challenge is a routine with the variables already fixed. A fresh challenge is generated at 00:00 UTC from a published seed, and every player in the world gets the same words: a 60-second English run, no punctuation, no numbers. That makes it the most directly comparable measurement the site offers, because for one day nobody is being scored on easier text than anybody else.
It also gives you an adherence metric that is harder to argue with than intention. Keyboardo tracks your current streak and your best streak of consecutive daily challenges, and the achievement catalogue rewards seven in a row, thirty in a row and thirty completed in total. Slot the challenge in as the measured test at the end of your loop and the whole routine has a fixed shape, a fixed condition and a visible record, with no decisions left to make while you are tired.
Tracking progress without fooling yourself
One test result is noise. Your score moves by several WPM between runs on identical material for reasons that have nothing to do with skill — how awake you are, which words came up, whether you were interrupted. Improving by three WPM once means nothing, and so does getting worse by three.
Every progress view on the site is therefore an average or a series rather than a single number. A profile shows the average WPM of your last ten tests, a chart of your daily average over the past 90 days, an activity heatmap of how many tests you completed on each of those days, and your personal best on each of the eight time and word boards. Read the heatmap and the chart together: a flat curve across a month of empty heatmap squares is not a plateau, it is a month you did not practise, and the two problems have entirely different fixes.
None of this requires signing up to begin. Your most recent 200 results are kept in your browser with no account at all, and if you create one later they can be merged in, so nothing you record while deciding is wasted. To start the loop now, take a test; for the technique the focused block is meant to build, see the touch typing guide and what actually makes people type faster.