How Long It Takes to Learn Touch Typing as an Adult: What Can Actually Be Counted

You have decided to fix your typing, and before you start you want to know what it costs. Not in money. In weeks.

So you search, and every result answers immediately. Four to six weeks. Two to three weeks with focused practice. Ten to fifteen hours. The numbers are confident, they are specific, and not one of the pages carrying them says where they came from.

I have looked. The first page of that search is almost entirely blogs run by typing apps — Mochi Type, the app I make, is now one more entry in that list, which is exactly why this page starts by telling you what it cannot tell you.

Why every page you have read says four to six weeks

Because the number is cheap and nobody checks it.

Look at who is answering. Typing Pal’s blog, Typiq’s blog, TypingFirst, TypingFlo, a handful of smaller apps, and a few personal essays. Eight or so near-identical pages, most of them run by a product that would like you to start today. None of them shows a study, a dataset or a cohort. The number is repeated because the number is already there.

The honest version is duller. How long it takes depends on three things nobody publishing a headline number knows about you:

• Minutes a day, actually done. Twenty minutes daily and forty minutes on Sunday are not the same programme, and the second one is what most people mean by “an hour a week”.
• What you are starting from. Learning a skill you do not have is a different job from overwriting a habit you have run at speed for two decades. The second one is harder and it feels worse.
• What counts as finished. Covering the whole keyboard without looking is one milestone. Getting back to the speed you already had is a second. Passing your old speed is a third. They can be months apart, and every blog headline blurs them into one word.

I could add a fourth variable — age — because you searched for this as an adult and you are probably wondering whether you left it too late. I have nothing measured to offer there, so I will not pretend. What I can say is that nothing in the mechanics of this changes with age: the course is the same list of keys in the same order, and the threshold does not care when you were born.

And the thing I most want on the record: Mochi Type contains no measurement of calendar time to fluency. It records sessions, per-key accuracy and personal bests, all of it on your own Mac — its privacy manifest declares no collected data types at all, so there is no dataset here for me to mine even if I wanted one. Every timeline on this page is somebody else’s anecdote, and it is labelled as such where it appears.

The part that can be counted is the syllabus

“Week three” is a metaphor. “Stage seven” is a list of letters. Here is the whole English course, in order, exactly as the app runs it:

Stage	What it adds
1	e, t, n, o
2	a, h
3	s, i
4	r, l
5	d, u
6	c, m
7	w, y
8	f, p
9	g, k
10	b, v
11	j, x, q, z
12	Capitals — Shift plus every letter you have learned
13	Punctuation — . , ! ?
14	Quotes — ' "
15	Extra Symbols — ; : ( ) -
16	Numbers — 0 to 9
—	Final Test

Sixteen stages and a test. Note what stage one is not: it is not the home row. asdf jkl; is the traditional opening and it is a hand-position argument, not a language argument — it puts you on keys you can reach rather than keys you will use. This course opens on the four most frequent letters of the language instead, chosen so that two fall under each hand, which is why you are typing note, tone and not in the first minute rather than fff jjj.

The same logic runs the whole way down. The order is that language’s letter frequency, hand-balanced against that language’s default layout — and the balance is checked by a validator that reports both word coverage and left-right split, because a stage that unlocks no real words is a stage of nonsense syllables.

Two more countable things about what appears on screen:

Word length is capped early and then released. Stages 1 to 3 cap words at five characters, stages 4 and 5 at six, stages 6 and 7 at seven, stages 8 and 9 at eight. From stage 10 there is no cap at all. So the text visibly grows up as you go, and it does it on a schedule rather than at random.

The new letters are forced into the text. The generator runs to a quota: 65% of the words in a stage’s text have to contain that stage’s new letter, and once a letter has been flagged as weak for you, 15% have to carry that one. Each new letter also gets its own shuffle bag, which is the part that matters for a rare letter: without it, z would drown in a pool of words picked by frequency, and you would barely meet it in a minute of practice.

Progress is a boolean, not a feeling

Most tutors keep their advancement rule to themselves. This is the whole of Mochi Type’s, from the code that decides it. A stage unlocks the next one when, inside a single 60-second session:

• accuracy is at least 95%, and
• speed is at least 25 WPM, and
• at least 10 keystrokes were recorded, and
• you are on the newest stage you have reached — replaying stage 4 for a better score cannot unlock stage 12 — and
• the course is not already finished.

All five, or nothing happens. And because a 60-second Learn session is the only unit of measurement here, the answer to “am I making progress” is available every single minute you practise, which is a considerably better instrument than a calendar.

There is one escape hatch, and it exists for exactly the person this page is written for. Accuracy is the hard requirement; speed is the one that stalls beginners for weeks. So: clear 95% accuracy but miss 25 WPM five sessions in a row, on the same stage, and the next stage opens anyway. One session below 95% resets that counter to zero. You cannot be trapped by slowness alone, only by inaccuracy.

The star rating is wired to the same number, deliberately. Three stars require 95% accuracy — the same 95% the gate wants — so “three stars” and “I can move on” are one statement rather than two. Above that: 96% is three and a half, 98% is four, and 100% accuracy is a flat five regardless of how slowly you typed it. A 50+ WPM session adds half a star, capped at five. That gives you a cheap nightly reading. Not “am I on track for week four”, but “did tonight clear the bar”.

Why the end of the course is slower than the beginning

Because the scoring changes under you, on purpose.

On the five advanced stages — Capitals, Punctuation, Quotes, Extra Symbols and Numbers — the accuracy that the gate reads is not your overall accuracy. It is a 50/50 blend of your overall accuracy and your accuracy on that stage’s own characters alone. On the Numbers stage, half of your score is carried by the digits. On Capitals, half of it is carried by the capitals.

This is the part people feel as “I was flying and then I hit a wall”. You were flying because letters are the bulk of any text and your letters had become good. The moment the stage’s own characters carry half the score, a clean letter run stops covering a fumbled reach for 4 or ;. Those are also the exact keys that a self-taught typist has never once hit without looking.

Budget for it. If somebody’s timeline told you that eleven letter groups is the whole job, the last five stages are the part they forgot to count.

The finish line has a specification

The Final Test is not vibes either. It is exactly 90 seconds in every build, run on the top 1,000 words of your language with punctuation, capitals and numbers all switched on at once. Three verdicts:

Verdict	Requirement
Pass	30+ WPM and 97%+ accuracy
Honors	40+ WPM and 98%+ accuracy
Perfect	40+ WPM and 100% accuracy

It does not become available until the Numbers stage — the last one — has been passed at 95%. That gate was added because merely reaching the last stage was not the same as clearing it: a sloppy Numbers run at 88% used to open graduation.

Those thresholds are worth reading as a target, because they are the only place in this whole page where a number attaches to a finish. Thirty words a minute at 97% accuracy is not fast. It is the point at which the keyboard has stopped being the thing you are thinking about, which is the actual goal, and everything past it is a matter of practice hours.

The hours other people report

Here is the anecdote, in full, because the truncated version of it is the whole internet’s favourite number and the ending is where the honesty lives.

From the Hacker News thread Learn touch typing – it’s worth it (May 2025), the most-upvoted comment:

I somehow managed to get through over a decade as a programmer before deciding to actually learn how to touch type. I was stuck at about 60 wpm with the wrong fingering, decided to invest in learning the right approach and dropped down to 30 wpm for a while, but eventually ended up at 120 wpm after a year or so.

One person. One report. No log, no methodology, no controls. It is evidence of nothing in the statistical sense, and I am quoting it anyway because it is the only thing on this page that answers the question you typed, and because it is honest about the two parts every marketing timeline drops: the drop happened, from 60 to 30, and the payoff was measured in a year, not in four to six weeks.

Read the shape rather than the numbers. A halving of speed while the new fingering is installed. A long climb after it. An eventual ceiling well above the old one — which is the argument for doing this at all, and it is the argument that four-week timelines quietly destroy, because a person promised four weeks quits in week two when they are typing at half their old speed and the calendar says they are behind.

If you are switching layout rather than fingering — QWERTY to Colemak or Dvorak — that is a related question with a different answer and different anecdotes, and it deserves its own page rather than a paragraph here.

Where to practise as an adult

The other half of what people search for, and it is a real problem: most typing practice is written for eleven-year-olds. Cartoon mascots, confetti, a racing car that goes faster when you type faster. It is not that gamification is beneath you. It is that it makes the app impossible to open at a desk with other people around.

What an adult actually needs from practice material is that it be worth reading. In Mochi Type’s Practice mode that means frequency tiers — the top 200 words of a language, then 1,000, 5,000, 10,000 and 50,000 — where the tier you pick decides how ordinary the vocabulary is. Top 200 is the, and, because. Top 50,000 has words you will have to think about. All five tiers ship for all 21 languages. Beside them sit movie quotes and book quotes, which exist for seven languages, and a set built from the words people most often misspell, which covers ten and is a meaner drill than it sounds.

Then there is your own text: paste or import something you were going to read anyway and type that instead. The free tier holds two of your own texts, and the built-in library ships alongside them.

On the privacy question, since an adult practising typing at work tends to care: your history is a SQLite file on your own Mac, there is no leaderboard, no friends list and nothing shared with anybody. The one caveat is iCloud sync, which on a Mac signed into iCloud starts switched on — the offline and no-account claims, and how to check them yourself, get a page of their own.

And the answer that costs you nothing: the free tier is the first five stages of every course, the top 200 word set in all 21 languages, and every keyboard layout. Five stages is five letter groups — e t n o a h s i r l d u — which is enough to know within a week whether the whole business is going to hold your attention, before any money is involved.

What this course will not do for your timeline

The limits, in one place, because a page about how long something takes owes you the parts that will annoy you.

There is no placement test, no skip-ahead and no test-out. The stage navigation refuses any index above the highest stage you have unlocked. If you already type 60 WPM with three fingers, you begin at stage one and clear each stage at 95% and 25 WPM like everybody else. For that reader this is the single most important sentence on the page, and it is the reason the first five stages are free: find out whether you can stand it before paying.

Learn sessions are 60 seconds and you cannot change that. The other lengths — 15, 30, 60 and 120 seconds — live in Practice mode, and unlimited runs are offered only on your own imported texts. The fixed minute is what makes the unlock threshold a single comparable measurement rather than a moving target, but if you wanted to sit down for a twenty-minute course session, that is not on offer.

There are no goals, targets or deadlines in the app. Nowhere to enter “40 WPM by October”. The statistics do keep a streak of consecutive days practised, but that is a record of what happened, not a target handed to you.

The stage count is identical in all 21 languages. Eleven letter groups and the same five advanced stages, whichever course you pick. What changes between languages is which letters fall in which group, and on some languages a group holds three or five letters rather than two. Nobody’s course is longer.

And the keyboard under your hands is not part of the timeline. If you are wondering whether an external keyboard would speed this up, the alphanumeric block is the same on a MacBook and on a Magic Keyboard, and what actually needs retraining when you move between Apple keyboards is the modifier corner, not your letters.

So: stop counting weeks and start counting stages. Sixteen of them, a published bar for each, sixty seconds per attempt. That is a measurement you own, it updates tonight, and it does not require anybody — me included — to have made up a number about you.