How We Test Keyboards
Principles
Our methodology is built to answer one question: how does a keyboard perform, sound, and feel in real play, and what is the simplest path to make it better. We use repeatable procedures, publish our assumptions, and update methods as the market evolves.
Test environment and controls
- Controlled desk acoustics with noise floor measurement and documented mic distance and orientation
- Consistent keycap profile and desk mat for baseline tests, with notes when stock parts are used
- Firmware versions and software settings logged, including polling rate, debounce, and rapid trigger states
End to end latency
We measure total input latency, not just switch actuation. Depending on interface, we use microcontroller-based latency rigs and high-speed capture to time the path from key movement to on-screen event. We test wired USB, 2.4 GHz dongles, and Bluetooth separately at the highest stable polling rates available. We validate polling consistency and packet loss under CPU load to simulate real game conditions.
Key checks include:
- Polling stability across 125 to 8000 Hz where supported
- Debounce and scan behavior, including rapid trigger or analog hall effect activation and reset distance
- NKRO integrity and rollover under multi-key stress
Acoustics and feel
We capture sound with calibrated microphones and analyze amplitude, decay, and frequency distribution to characterize thock, clack, and ping. We test case resonance, plate material behavior, and mounting style. We document stabilizer rattle and wire tick. For each board we provide baseline samples and post-mod samples so you can hear expected changes.
Ergonomics and fit
We record front height, typing angle, optional riser angles, and wrist support needs. We evaluate layout efficiency for play and productivity, including access to F keys, nav clusters, and macro layers. We test comfort over long sessions with different hand sizes and note fatigue risks and mitigations.
Mods and upgrade paths
Every recommendation includes a step-by-step recipe with parts lists:
- Switches: options by force curve, travel, and sound; lube suggestions and film compatibility
- Stabilizers: tuning and wire treatment to reduce rattle and tick
- Case and plate damping: foam types, tape mods, and when to avoid over-damping that blunts feedback
- Firmware and macros: remaps, layers, and on-board storage to avoid heavy background drivers
We grade mods by cost, time, and impact so you can choose quick wins or deeper builds.
Reliability, repairability, and future proofing
We inspect solder quality, hot-swap socket tolerance, and connector strain relief. We assess spare part availability, standard layout compatibility, and teardown difficulty. Firmware update safety and rollback are noted.
Software testing
We favor stable, driverless workflows. Where vendor software is required, we test for system impact, offline operation, macro reliability, and export or import of profiles. Open firmware support like QMK or VIA is highlighted when robust.
Scoring
Scores weight what affects play first: latency and stability, then acoustics and ergonomics, then software and build. We publish the weights and show how mods can move a board between upgrade tiers. When two boards tie, we break ties by value per dollar and quality of the long-term path.
Repeatability and uncertainty
Each test includes multiple runs and the results are averaged with outlier checks. We report notable variance and confidence ranges where relevant. If a firmware update changes behavior, we retest and annotate the review.
What we do not do
We do not score based on brand hype, RGB quantity, or fleeting trends. Looks matter, but not at the expense of performance, comfort, or reliability.
Continuous improvement
We audit our methods quarterly, accept community challenges with data, and publish changes to procedures so our results remain transparent and useful.