GuidesComputing

Mechanical Keyboard Switches, Explained Without the Marketing

Linear, tactile and clicky are simpler than they sound, and the parts nobody advertises decide whether a keyboard feels expensive or hollow.

July 18, 2026 · 14 min read · 3,123 words

Linear, tactile and clicky: the entire taxonomy

Under every keycap sits a self-contained switch: a plastic stem riding on a spring inside a housing, with a metal leaf or a contactless sensor that closes the circuit as the stem descends. Behaviour splits three ways and no further. A linear switch slides down with resistance that climbs steadily and nothing interrupting it. A tactile switch adds a bump partway through the press, a short stretch where force rises and then drops away. A clicky switch takes that same bump and bolts on a separate part that snaps to make noise. Colour names sit on top of all this as factory branding rather than an industry standard.

Feedback from the bump only pays off if you act on it. Plenty of typists press straight through and drive the key into the plate anyway, which means paying for a sensation they immediately overrule. Heavy typists therefore find linear and tactile switches converge until the difference is close to academic. Light typists, the ones who release as soon as a key registers, get something concrete out of it: the bump reports that the work is done, and stopping there across eight hours removes a real amount of accumulated impact from your fingertips and wrists.

The click deserves plain description. A small collar or a sprung bar inside the housing releases under load and strikes the housing wall, and that noise is the entire purpose of the part. Registration happens at the same depth it would in a quiet switch, so nothing about the click improves speed or accuracy. It also has a quirk that rarely gets mentioned. The clicking mechanism and the electrical contact are separate events tuned to coincide on the way down, and they do not coincide on the way up, which is why fast double taps on a clicky board can sound out of step with what registers.

Within the tactile family, bump placement varies far more than the labels admit. Some designs put the bump right at the top of the stroke, where it feels sharp and almost aggressive, with the key registering immediately afterwards. Others place a rounder, gentler bump nearer the middle, followed by a stretch of smooth travel. Those two feel nothing alike despite sharing a category. When a listing includes a force curve graph, read where the peak sits relative to the actuation point. That single detail predicts how a switch feels better than the type name ever does.

Force and travel, and the numbers that mislead

Spec sheets quote actuation force in grams, which is shorthand for gram-force; some list the same figure in centinewtons, where 45 grams is roughly 4.4 cN. Most switches sold for typing sit between about 45 and 65 gram-force at the point of actuation. That range looks meaningful in a table and reads narrower under your hand. Ten grams either way is obvious in a deliberate side-by-side comparison and largely invisible once you are typing sentences. What registers over hours is the shape of the curve: whether resistance climbs gently, spikes early, or ramps hard in the final millimetre so the switch seems to push back.

One number gets quoted and a second one, usually absent, describes your day better. Actuation force is measured where the key registers. Bottom-out force is what your finger meets at the end of travel, typically 15 to 25 gram-force higher because the spring has compressed further. Since most people bottom out on most keys, that second figure is the one you feel. Progressive and two-stage springs widen the gap deliberately, staying light near the top so keys trigger easily, then stiffening towards the base so you stop hammering the plate. A switch marketed as light can still finish heavy.

Travel is how far the key moves before it stops. A conventional switch runs about four millimetres in total and registers near two. Low-profile designs cut that to roughly three millimetres with actuation around 1.2, which feels quick and suits anyone arriving from a laptop, though the shrunken margin means fingers resting on the home row trigger keys more easily. Competitive players sometimes want the shallower actuation point for faster repeated presses. Writers usually prefer the depth, since one stray character costs more attention than a millisecond of latency ever saves back.

Treat very light springs with caution at the start. At 35 gram-force and below, the weight of a hand resting on the home row is enough to fire keys until you retrain your posture, and that retraining runs to weeks rather than days. At the other end, springs in the 60 to 70 range feel deliberate and controlled, and they tire some people during long writing sessions. Landing mid-range is the low-risk opening move. Springs are also the cheapest thing in a keyboard to change later, so the decision commits you to very little.

This is the dominant source of typing noise and most of the impact your fingers absorb, and it is a habit rather than a specification. It also explains why bottom-out force, usually 15 to 25 gram-force above the advertised actuation force, describes how a switch feels better than the number on the box does.

On Bottoming out

Hot-swap sockets, and whether you will really use them

A hot-swap board carries small sprung sockets soldered to the circuit board, so a switch pulls out and pushes in by hand. The alternative has switches soldered directly to the board, which is permanent unless you own an iron, a desoldering pump and a free afternoon. The convenience is real, and the feature has spread from enthusiast boards down into ordinary ones, so a small premium for it is easy to justify. It rescues a keyboard from a single dead switch, and it lets you change the feel of a board you otherwise like without replacing the whole thing.

Be honest about the frequency. Most owners swap once, occasionally twice, then leave the board alone for years. Repair is where the feature earns its place. One failed switch on a soldered board means living with a dead key, remapping around it, or dismantling the assembly to heat two joints. On a hot-swap board the same failure is a ninety-second job with a two-pronged puller, and spare switches are sold in bags of ten for very little. On a keyboard you intend to keep for a decade, that argument settles it on its own.

Two compatibility traps catch people out. Switches have either three or five pins underneath, the extra two being plastic alignment legs, and some boards accept only the three-pin kind. Five-pin switches can be trimmed with flush cutters, which takes seconds and cannot be undone. The larger trap is technology rather than shape. Sockets built for standard mechanical switches will not take optical or Hall effect switches, and the reverse holds too, because those register presses by completely different means. A board advertised as hot-swappable is hot-swappable only within its own family.

Sockets vary in quality and none of them are indestructible. The usual cause of a permanently dead key position is levering a switch out at an angle instead of straight up, which drags a socket contact sideways until it loosens from the board. Support the board from underneath, grip both housing clips evenly, pull vertically. Inspect pins before insertion as well. A bent leg forced into a socket splays the contact from the inside, and while a leg straightened once with tweezers is usually fine, a leg straightened twice tends to snap off in the socket.

Stabilisers separate good boards from cheap ones

Stabilisers are the wire and plastic assemblies fitted under keys wider than about two units: space bar, both shifts, enter, backspace, and often the tall keys on a number pad. A stiff wire links two plastic housings so pressing one end of a long keycap lowers the whole cap evenly instead of tipping it. They are also the most common reason a keyboard sounds cheap. A loose stabiliser wire ticks against its housing on every press, and that noise is metallic and hollow, nothing like the switches around it. No switch upgrade covers it up.

This is the fastest quality test available to a buyer, and it works in a shop or inside a return window at home. Skip the letter keys, because every keyboard feels acceptable on the letters. Press the space bar at the extreme left, the extreme right, then the centre. Repeat on right shift, enter and backspace. You are listening for the same sound in all three positions. A clatter at one end against a clean thud in the middle means the wire is loose or dry. Two keyboards compared this way separate themselves in about fifteen seconds.

Stabilisers respond well to cheap intervention. A thick dielectric grease applied where the wire ends pivot inside the housings kills most of the ticking. A thin strip of fabric plaster or tape laid on the board beneath the housing softens the impact of the wire returning, which accounts for the other half of the noise. Both fixes are well documented and cost almost nothing beyond an hour of your time. Whether you want to pull keycaps, and possibly unscrew the board to reach the underside, is the question that actually decides it.

If the answer is no, treat stabiliser quality as a hard purchase criterion, and learn the two mounting types before you shop. Screw-in stabilisers bolt through the circuit board and sit rigidly, and they are what better boards use. Clip-in and plate-mounted stabilisers snap into the plate, cost less to fit, and wobble more. A listing that specifies screw-in stabilisers is telling you something honest about where the budget went. A listing that says nothing on the subject generally has nothing worth advertising.

The case and plate shape the sound more than the switch does

Switches take the credit for a keyboard's character while contributing less to it than the structure around them. The plate is the sheet switches clip into, and its material sets the stiffness of every press. Steel is rigid and bright. Aluminium sits in the middle and is the common default. Brass is denser and deeper. Polycarbonate flexes under the finger and gives a softer, lower tone. Some designs omit the plate entirely and clip switches straight to the circuit board, which is softer still and slightly less stable laterally. One switch across two of these arrangements sounds like two different products.

The case does more still. A thin plastic shell with a large cavity beneath the board behaves like a resonating chamber, which is why inexpensive keyboards sound thin and echoing whatever is fitted inside them. Filling that void changes the result immediately. A sheet of foam between the board and the case floor absorbs the reflection, a second layer between board and plate deadens the switch housings, and heavier cases resonate less to begin with. This is also the cheapest worthwhile modification anyone makes: an offcut of craft foam cut roughly to shape gets most of the way there.

Mounting style decides how the whole assembly moves. In the cheapest arrangement the board screws to posts moulded into the case bottom, which is rigid, uneven, and produces harder-sounding keys in the areas nearest the screws. Top mount hangs the plate from the upper shell instead. Gasket mount suspends the plate on strips of silicone or foam so the typing surface flexes slightly under pressure and absorbs energy rather than transmitting it into the case. That cushioned quality is usually what people are chasing when they buy a fourth set of switches for a board that was never going to provide it.

There is a sensible order to spending, and it follows directly from all this. Choose the keyboard first on case construction, plate, stabilisers and layout, because not one of those can be changed without buying another keyboard. Choose switches second, since a full set is the least expensive component to replace and takes an evening to fit. Reversing the order produces the familiar outcome: immaculate hand-picked switches sitting inside a shell that will never sound good no matter what goes into it.

Noise in shared rooms, and what your colleagues hear

Keyboard noise has two sources and buyers routinely conflate them. The first is the switch mechanism: the click if there is one, plus the spring and the leaf. The second is bottoming out, meaning the keycap and stem arriving at the housing and the plate at whatever speed your finger delivered them. For most typists the second is several times louder than the first. Someone running silent linears can still be the loudest person on the floor, because technique sets the volume far more than the part number does.

For shared rooms the hierarchy is short. Clicky switches do not belong in an open office, and enthusiasm on your side does not change the experience on theirs. Ordinary linears and tactiles sit on the borderline and depend almost entirely on the case and on how hard you type. Silent switches are the genuine fix: rubber or silicone pads moulded onto the stem cushion both the downstroke and the return, cutting impact noise substantially. The trade is a slightly mushy, less defined bottom, and on silent tactiles the damping can blunt the bump too. Most people who need the quiet accept both.

Cheap mitigations punch well above their weight here. A dense desk mat removes the reflection off a hard desk surface, which contributes more than its price suggests. Thicker keycaps absorb impact instead of ringing. Rubber O-rings slipped onto keycap stems shorten travel by a fraction of a millimetre and soften the landing, though they change the feel enough that you should fit a handful of keys and live with them before doing the whole board. Typing with less force costs nothing at all and beats every item on this list.

Video calls are the harsher test. A laptop microphone or a headset boom sits far closer to the desk than any colleague does, and it captures both the impact noise and the low thump conducted through the desk surface itself. If calls are the actual problem rather than the room, move the keyboard onto a separate surface from the microphone, raise the mic, or use push-to-talk. Any of those does more for the people listening than a switch swap, and none of them requires opening the keyboard.

The move from the old 125 hertz standard, one report every 8 milliseconds, to 1000 hertz at 1 millisecond was a genuine improvement. Above that the arithmetic stops mattering: 8000 hertz shaves 0.875 milliseconds, less than the frame interval of even a 240 hertz display, while consuming measurably more CPU on some systems.

On Polling rate

Keycaps, and the specification listings hide

Keycaps affect sound and feel more than most buyers expect, and material is the first fork in the road. ABS is smoother from new and develops a shiny, greasy-looking patch where fingers land, often within a year of daily use. PBT has a faintly textured surface, resists that polish for years, and sounds slightly deeper because the plastic is denser and stiffer. Wall thickness matters alongside material: caps around 1.5 millimetres sound fuller and more solid than the 1.0 to 1.2 millimetre caps shipped on most prebuilt boards. Thick PBT is the sensible default now the price gap has closed.

Profile describes the height and shape of the caps row by row. Sculpted sets angle and raise each row differently so the keyboard meets your fingers where they naturally sit. Uniform sets make every row identical, which photographs beautifully and feels slightly wrong for about a week. Height varies independently of that: tall, deeply dished profiles feel enveloping and are noticeably louder, while short flat profiles feel quick and stay quieter. Changing profile resets some muscle memory temporarily, which is worth knowing before ordering a set chosen mainly on how it looked in a product photo.

Legend printing is the specification that catches people out most often. Pad printed legends sit on top of the plastic and rub off, sometimes within months on the keys you use hardest. Laser etching lasts longer and tends to collect grime in the recess. Double-shot caps mould the legend as a separate piece of plastic running through the cap, so the character cannot wear away, although the surface around it can still shine. Dye sublimation dyes the legend into the plastic and is similarly permanent, with one restriction: the dye must be darker than the cap, so light legends on dark caps are made another way.

Compatibility is the final trap. Most sets are cut for a cross-shaped stem, and switches using a different stem shape, including several optical designs, need caps made for them. Layout catches more people than stems do. Sets are sized for a standard bottom row, and boards with an unusual space bar width or oddly sized modifier keys will leave gaps a normal kit cannot fill. Count the key widths on your bottom row against the kit contents before ordering. That mismatch is the single most common reason an expensive keycap set ends up resold.

Putting it in order

Settle the keyboard before the switches, because layout and construction govern your posture and your sound every day and cannot be changed afterwards. Full-size includes the number pad and pushes your mouse hand outward, which loads the shoulder over long sessions. Tenkeyless drops the pad and brings the mouse back towards your body line, an ergonomic gain that gets underplayed in reviews. Compact layouts move the function row and sometimes the arrows onto a held layer, which looks elegant and irritates anyone who reaches for those keys without thinking about it first.

Then start with the room, since it removes options faster than anything else. Sharing a space rules out clicky switches and pushes you towards silent variants, and the rest of the decision narrows sharply once that is fixed. Working alone leaves the field open, at which point feel is a perfectly legitimate basis for choosing. Look next at what you spend hours doing. Long-form writing rewards a tactile bump and a mid-weight spring, both of which discourage bottoming out. Gaming leans linear, mainly for rapid repeated presses, though the margin is smaller than the advertising implies.

A switch tester is worth buying if you remain undecided. These are small blocks holding a handful of switches, cheap enough to treat as disposable. They are also systematically misleading about sound, because a switch sitting in an unfilled plastic frame has none of the plate stiffness or case resonance of a finished board. Use one to judge spring weight, the position and sharpness of any bump, and how smooth the downstroke feels. Judge sound instead from recordings of the specific keyboard you are considering, and listen to several from different people before trusting any of them.

Then stop optimising. The distance between a competently built board with mid-weight switches and an obsessively tuned one is audible and sits well past the point of diminishing returns. A solid case, screw-in stabilisers, thick PBT caps, hot-swap sockets and a switch weight suited to the room you work in will cover almost everything a person notices day to day. What lies beyond that is a hobby, and an enjoyable one, but it is not a prerequisite for typing comfortably for the next ten years.

What to avoid

Buying premium switches for a board with a hollow plastic case. The case, plate and mounting determine most of what you hear, and excellent switches inside a resonant shell still sound thin. Add foam to the cavity or accept the sound, but do not expect switches to fix it.
Choosing switches by colour name across manufacturers. Colours are factory branding rather than a standard, and a given colour from one supplier can be lighter, heavier or a different family entirely from the same colour elsewhere. Read the type and the force rating, and ignore the name.
Paying extra for polling rates above 1000 hertz, additional RGB lighting zones, or per-key macro software you will configure once and never reopen. None of these alter how the keyboard feels under your fingers, which is the only property you still notice after the first week.
Assuming silent switches make you silent. They damp the switch mechanism, not the sound of keycaps being driven into the plate by a heavy hand. If you type hard, a dense desk mat, thicker keycaps and a lighter touch will do more for the room than any switch swap.
Skipping the stabiliser check because the letter keys felt good in the shop. Every keyboard feels fine on the letters. Press the space bar, right shift and enter at both ends and the middle, and buy the board where all three positions sound the same.

Common questions

Are mechanical keyboards genuinely better than membrane ones, or is it enthusiast folklore?: Better is the wrong frame for it. Mechanical switches are consistent across the board, more durable, and repairable, because each key has its own independent mechanism instead of sharing anything with its neighbours. Membrane keyboards press a single rubber sheet against a printed circuit, which gives a mushier press that varies from key to key and softens further as the rubber ages. Whether you care depends on how much you type. Someone producing a few thousand words a day notices inside a week. Someone who mostly drives a mouse may never notice, and that is a reasonable place to land rather than a failure of taste.
How long do mechanical switches last in practice?: Manufacturers quote 50 to 100 million presses per switch, which sounds absurd and is roughly accurate for the contact mechanism under lab conditions. Switches rarely die of wear in the real world. They fail from contamination, meaning dust, crumbs and spilled drinks reaching the leaf, or from a spring or stem fault that was present from the factory and took months to show. This is the strongest practical case for hot-swap sockets. You are not planning for gradual decay across the whole board, you are planning for one random switch failing on an otherwise perfect keyboard four years from now.
Should I lubricate my switches myself?: Not as a starting point. Hand lubricating a full board means opening every switch, brushing a controlled amount onto specific rails and the spring, and reassembling all of it. Budget several hours for a first attempt, and expect an improvement that is real but subtle. Decent factory lubed switches capture most of that benefit for none of the labour. If you want to lubricate exactly one thing, make it the stabilisers. There are five or six assemblies rather than a hundred switches, the grease goes on the wire ends where they pivot, and the change in sound is far more audible than anything you will achieve inside the switches.
Is a tactile bump the same thing as a click?: No, and conflating the two causes more bad purchases than any other misunderstanding. A tactile bump is felt, produced by the profile of the stem pushing past a sprung metal leaf. A click is heard, produced by an extra moving part whose only job is to make noise against the housing. Every clicky switch is also tactile, since the clicking mechanism generates a bump as a by-product. The reverse does not hold: plenty of tactile switches are as quiet as linears. If you want the feedback without announcing yourself to the room, quiet tactiles are common, cheap and easy to find.
Can I make a keyboard I already own sound better without replacing it?: Usually yes, and in a fixed order. Start with the stabilisers, because grease on the wire ends and a strip of tape under the housings removes the rattle that dominates the sound. Next, open the case and lay foam in the empty cavity beneath the circuit board, which kills the hollow echo that makes cheap boards sound like a drum. Then try thicker PBT keycaps, which absorb impact rather than ringing. Put the keyboard on a dense desk mat while you are at it. Switches come last, and by that point many people find they no longer want to change them.

Computing, decoded

Struck = safe to ignore= what it really tells you

Actuation point

How far down the key travels before the keystroke registers, measured in millimetres from the top of the stroke.

Relevant to competitive gaming, where rapid repeated presses benefit from a shorter distance and a higher reset point. Nearly irrelevant to typing, where a shallow actuation point mainly buys you accidental keystrokes. Adjustable actuation on Hall effect boards is a real feature for a small number of players and an unused menu item for everyone else.

Bottoming out

Driving a key all the way to its stop rather than releasing at the depth where the keystroke registers.

This is the dominant source of typing noise and most of the impact your fingers absorb, and it is a habit rather than a specification. It also explains why bottom-out force, usually 15 to 25 gram-force above the advertised actuation force, describes how a switch feels better than the number on the box does.

N-key rollover

The keyboard reports every key held down at once without dropping any of them.

Useful in a handful of games with dense simultaneous inputs and essentially never relevant to typing. Nearly every wired mechanical keyboard has it now, so it is not worth paying for. The exception nobody tests: many boards fall back to six-key rollover over Bluetooth and inside a computer's pre-boot firmware menus.

Polling rate

How many times per second the keyboard reports its state to the computer, quoted in hertz.

The move from the old 125 hertz standard, one report every 8 milliseconds, to 1000 hertz at 1 millisecond was a genuine improvement. Above that the arithmetic stops mattering: 8000 hertz shaves 0.875 milliseconds, less than the frame interval of even a 240 hertz display, while consuming measurably more CPU on some systems.

Factory lubed switches

The housing rails and stem have been coated with a thin lubricant during manufacture to cut friction and scratchiness.

Improves smoothness noticeably and removes the gritty feel of budget switches. Quality varies enormously between factories, and heavy application can smear a tactile bump into vagueness or slow the spring return. On linears more lubricant is generally welcome; on tactiles it frequently is not.

Optical and Hall effect switches

Switches that register a press with a light beam or a magnetic field instead of two metal contacts meeting.

Removing physical contact eliminates contact bounce, so firmware can skip the few milliseconds of debounce filtering metal contacts require, and it removes a wear point, making durability claims fair. The latency gain is a fraction of a millisecond and will not change results. Hall effect reads stem position continuously, which is the real draw: adjustable actuation depth and instant reset. Both types need matching boards and sometimes matching keycaps.

The shelf

Discounted in Computing

Products in this category our engine currently flags as below their usual list price. Apply the guide above before you buy any of them.