Keyboard latency
Keyboard latency marketing focuses on polling rate, which is the smallest term. The two that matter are debounce delay, which firmware chooses, and travel distance, which your finger has to physically cross.
The chain
| Stage | Typical cost | What determines it | Can you change it? |
|---|---|---|---|
| Travel to actuation | 5 to 14 ms | Actuation depth and your finger speed | Yes: shallower actuation, shorter throw switches |
| Debounce | 0 to 8 ms | Switch technology and firmware setting | Yes: optical or Hall effect, or a lower firmware setting |
| Matrix scan | 0.1 to 1 ms | Board firmware scan rate | Rarely exposed |
| Report wait | 0.5 ms at 1000 Hz | Polling rate | Yes, but the saving is tiny |
| Wireless hop | 0.2 to 1 ms | Radio protocol | Use the dongle, never Bluetooth |
The ordering is the whole point. Travel time and debounce together account for the large majority of the total, and both are addressable. Polling rate, which is the number printed on the box, is the smallest term in the table.
Debounce, and why it is a choice
Contact bounce is a physical property of metal switches, and firmware must wait it out. But the length of the wait is a firmware decision balancing latency against reliability: too short and a bouncing switch registers extra presses, too long and every keypress is delayed.
Manufacturers pick conservative values because a keyboard that types double letters is a returned keyboard. That conservatism costs you milliseconds. Some boards expose the setting; if yours does, lowering it from 8 ms to 3 ms is free latency, and if you start seeing doubled characters you have gone too far and can step back up.
Optical and Hall effect switches have no contacts, so there is nothing to bounce and the term is genuinely zero. This is their most concrete latency advantage and it is rarely how they are marketed.
Release latency is the one that matters
Almost all keyboard latency discussion is about how fast a press registers. In competitive play the release is more often the decisive event, because stopping accurately depends on it.
Release latency on a conventional switch has an additional term that press latency does not: the distance from wherever you pressed to back up past the reset point. If you bottomed out, that is over 2 mm of lift, which at typical finger speeds is more than 10 ms, and it varies with how hard you pressed.
This is why rapid trigger produces a bigger practical improvement than any latency specification on a box. It is not shaving a millisecond off a press; it is removing a variable ten-millisecond term from a release.
The fix order for keyboard latency
- Reduce actuation depth on movement keys, if your board allows it. Largest single saving.
- Get rapid trigger on movement keys, if available. Removes the variable release term.
- Lower debounce, if your board exposes it. Free milliseconds on conventional switches.
- Ensure 1000 Hz polling and a wired or 2.4 GHz connection.
- Stop. Everything beyond this is below the noise floor.
Related reading
- Actuation point, the largest term in the table above.
- Rapid trigger for the release problem.
- The full input lag budget, where keyboard and mouse both sit.
Frequently asked questions
What is keyboard debounce delay?
The period the firmware waits after detecting a contact change, to be sure the metal contacts have stopped bouncing before reporting a keypress. It is typically 3 to 8 ms on conventional mechanical boards and zero on optical and Hall effect boards, which have nothing to bounce.
Does keyboard polling rate matter?
Less than mouse polling rate, because keyboard input is discrete rather than continuous. At 1000 Hz the mean report wait is 0.5 ms, which is small next to debounce and tiny next to travel time. Going above 1000 Hz on a keyboard is not worth pursuing.
Which keyboards have the lowest latency?
Boards with no debounce requirement, meaning optical and Hall effect, combined with a shallow actuation point. The switch technology removes several milliseconds and the shallow actuation removes several more. On conventional mechanical boards, look for a configurable debounce setting.
Is wireless keyboard latency a problem?
A modern 2.4 GHz keyboard adds well under a millisecond, which is negligible against debounce and travel. Bluetooth is a different matter and is unsuitable for competitive play for the same reasons as on a mouse.
Does a faster keyboard actually win me fights?
In the specific case of counter-strafing and rapid direction changes, yes, because those depend on precise release timing. For pressing an ability, a few milliseconds is far below the noise floor of your own reaction time variability. Prioritise release behaviour over headline latency.