A mechanical keyboard will not make you type much faster. Expect a few WPM at most, and probably none. It can make long sessions more comfortable and give you clearer feedback on every keystroke, which may help accuracy, but the speed comes from your hands and your practice, not from the switches.
The question shows up on every keyboard forum in roughly this form: "I'm at 65 WPM on a laptop. If I buy a mechanical keyboard with linear switches, how much faster will I get?" We'd rather say "barely" than sell you on the thing this whole article is about.
The longer answer is more interesting. Keyboards change how typing feels, how long you can do it comfortably and what the room sounds like. Mechanical keyboard typing speed, measured the way we measure it, mostly isn't one of the things that changes.
What a switch actually does
A mechanical switch has a stem that moves down inside a housing, a spring that pushes it back, and a contact that closes at some point in the travel. Manufacturers publish the numbers that describe the motion.
Actuation distance, which Cherry calls pre-travel, is how far the key moves before it registers. Cherry's MX2A Red registers at 2.0 mm out of 4.0 mm of total travel. Its Speed Silver variant registers at 1.2 mm out of 3.4 mm. Scissor switches, the kind in most laptops, usually travel between 1.0 and 2.5 mm in total, and the butterfly designs some laptops used go below 1 mm.
Operating force is how hard you press to get there. Both of those Cherry switches need 45 cN, about 46 grams-force. Lighter and heavier springs are common.
Feel is the shape of the force curve. Linear switches have a smooth curve with no event at the actuation point. Tactile switches have a bump just before actuation, so your finger feels where the key registers. Clicky switches add a mechanism that makes an audible click at the same point.
That's the whole taxonomy. Everything else is tuning: spring weight, stem material, lubrication and the keycap profile on top.
Why actuation distance barely matters for speed
A typist at 80 WPM makes about 400 keystrokes a minute, one every 150 milliseconds. Dhakal et al.'s 2018 study of 136 million keystrokes from 168,960 volunteers measured an average inter-key interval of about 239 ms. Suppose a shallower switch saved 0.8 mm of travel per keystroke, roughly the gap between those two Cherry switches, and suppose a finger moves through that travel at a modest 100 mm per second. That's 8 ms per keystroke, out of 150 to 239. Three to five percent, and only if every keystroke saved the full amount. It doesn't, because fast typists don't wait for one key to come back up before starting the next.
The study's timing data makes the point from the other side. What separates fast typists from slow ones is rollover: pressing the next key before releasing the previous one. The mean rollover ratio across all volunteers was 25%, and most fast typists rolled over on 40 to 70% of keypresses. When rollover happened, keystrokes overlapped by 30 ms on average. That's a property of the typist's hands, and of the keyboard's ability to register overlapping presses, not of how far the keys travel.
So the only hardware property with a visible path to speed is the next one.
Rollover, NKRO and ghosting
A keyboard has to register the next key while the last one is still down. Nearly all do for two keys. Some cheap membrane keyboards and laptops use a key matrix that can misread certain three-key combinations: a key is dropped, or a phantom key appears where none was pressed, which is called ghosting. Mechanical keyboards generally advertise N-key rollover (NKRO), meaning every key is detected independently, or at least the six simultaneous keys of the standard USB boot protocol, which is more than enough for typing.
For a typist, the practical test is whether fast rolls on common pairs ("th", "er", "in") ever drop a letter. If they do, that's one real reason to replace the keyboard: a dropped keystroke is an uncorrected error you didn't make.
Rollover matters to us for another reason. Our anti-cheat checks look at timing features, including your rollover ratio, and fast humans almost always overlap keys. Your keyboard's rollover is, in a small way, evidence that you're a person. How we detect cheating has the detail.
Linear vs tactile vs clicky switches for typing
What we've read, and what we've seen in the per-key stats of the few of us who switch keyboards a lot, points to differences in error rate and comfort rather than speed.
Tactile switches tend to reduce bottoming out, because the bump tells your finger the key registered and you can lift. Over a long session that means less force per keystroke and less fatigue. Some find the bump interrupts their rhythm; we know of no controlled data showing either group is faster.
Linear switches feel smooth and have a reputation as the fast choice among people who type for sport. If that reputation is accurate, it's probably selection: competitive typists prefer a stroke with no interruption, and they'd be fast on either.
Clicky switches give an unmistakable signal that the key registered. For learning, especially the numbers and symbols far from the home row, that signal is useful, and it's the one case where we'd steer a learner toward clicky. In a shared office we'd steer you away, for reasons that have nothing to do with typing.
What does change error rate, in our experience, is keycap profile and spacing. Moving from a laptop to a mechanical board with roughly twice the travel and sculpted keycaps changes where your fingers land for a week or two. The most-missed keys map on the results screen lights up on the keys at the edge of your reach. It settles. It's also why we'd suggest not changing keyboards in the fortnight before an exam typing test.
Sound, and why we synthesize it
The sound of a keyboard matters more to people than they admit, and we built our sound packs around that.
The problem any typing sound in a browser has to solve is lag. A sound that arrives tens of milliseconds after the key is worse than silence, because your ear hears a rhythm your fingers didn't play. The arithmetic is unforgiving: at 80 WPM your keys land 150 ms apart, so a click that is 40 ms late sits more than a quarter of the way to the next keystroke. Some lag comes from the page, which we can fix, and some from wireless earbuds, which add their own delay no website can remove. If clicks feel late on a wireless headset, try the laptop speaker.
Our twelve packs (soft click, thock, tactile, linear, clicky, typewriter, pop, beep, pentatonic, piano, bubbles and marimba) are generated in the browser with the Web Audio API instead of played from recordings. There's no audio file to download or decode, so the first keystroke after the page loads makes a sound and so does the thousandth, with the same latency. The sound fires on keydown, not on the input event that follows it, because keydown is the earliest moment the browser knows a key went down. We retuned the gains in October after feedback that the packs were too quiet. The default volume now maps to a master gain of at least 0.45, every pack's loudest voice peaks at 0.6 or above before the limiter, and unit tests check the volume curve and every pack's peak so they can't drift.
The "thock" and "clicky" packs are modelled on what people describe when they talk about switches: a low, damped knock for the first and a sharper click with a high component for the second. They're approximations. A synthesized click will never sound like a lubed linear on a brass plate. The packs are for laptops and for people who want the feedback without the hardware.
The error sound is separate and we'd leave it on while learning: a muted low tone with a short noise knock, deliberately dull so a run of errors doesn't become a siren.
What actually changes typing speed
Here's the order we'd rank things in.
- Practice on your weak letter pairs. In the 2018 data, how quickly a typist moved between hands on alternating pairs tracked speed at r = -0.72. No keyboard changes that. The practice plan does.
- Accuracy. Net WPM counts only clean words, and every correction costs a backspace and a pause. The accuracy article has the arithmetic.
- Fingering. A finger assignment that worked at 40 WPM and caps you at 60 is more common than any hardware limit. The per-key stats show it as one key with a much higher interval than its neighbors.
- Rollover. If your keyboard drops keys on fast rolls, replace it. Otherwise it's fine.
- Everything else about the keyboard: switch type, actuation distance, keycap material, case. Comfort, feel, sound. Not speed.
That's why we'd tell the 65 WPM laptop typist to buy a keyboard if they want one, and to expect the number to move because of what they do on it.
What are the best switches for typing?
For a first mechanical keyboard, a tenkeyless or 75% board with a well-regarded tactile or linear switch on a light-to-medium spring, from a maker that publishes actuation specs and supports NKRO. Try switches before buying if you can. The difference between "I like tactile" and "I hate tactile" is a minute with a switch tester, and nobody can predict it for you. If you write code, consider a split or compact board with a symbol layer; that's a layout gain more than a switch gain, and our post on touch typing for programmers covers it.
If you don't want to spend anything, turn on the "tactile" pack here, switch the error sound on and take a typing test. The feedback is most of what the keyboard would give you, and the keyboard you have almost certainly handles rollover already. Run a fast "the the the" and check.
Frequently asked questions
Do mechanical keyboards improve typing accuracy?
They can, a little, mainly through clearer feedback and less fatigue on long sessions. Expect a week or two of extra errors while your fingers adjust to the new travel and keycap shape.
Is a lower actuation point faster?
Only by a few milliseconds per keystroke at most, against an average gap between keys of about 239 ms. Rollover technique matters far more.