By Eric, BlinkBench Founder · Last reviewed: July 2026
Most of the numbers on this site survive a change of device with a caveat attached. A reaction time is a reaction time whether you clicked or tapped, just measured through a different input path. Typing speed is different in a way worth spelling out, because the unit stays intact and the meaning doesn't — and that difference is the reason this site is built mobile-first rather than as a desktop test with a phone version bolted on.
The gap is not small
Two studies, run by the same research group with the same browser-based transcription task, give a clean before-and-after. A physical-keyboard sample of 168,960 people averaged 51.56 WPM (SD 20.2). A touchscreen sample of 37,370 people, recruited the same way and typing the same kind of sentences, averaged 36.2 WPM — a gap of about 15.4 words per minute, entirely attributable to the surface people typed on.
That is not two populations of different typing skill. It is the same task, on two different pieces of glass and plastic, producing two different averages. WPM did not stop being a valid unit — five characters is five characters regardless of what you typed them on. What changed is what a given WPM figure tells you about the person who produced it.
Touch isn't one thing either
The mobile study didn't stop at a single average. Over 74% of its participants typed with both thumbs, and that group was measurably faster than people using one thumb or a single index finger. People typing with only one or two fingers reached roughly 70% of the speed the desktop sample managed on a full keyboard. So “mobile” isn't a second fixed reference point next to “desktop” — it's a range of grips with a range of speeds inside it, and the device alone doesn't tell you which one a given visitor used.
This is why the typing speed test shows two reference figures rather than one range with a footnote, and infers which applies from pointer type rather than assuming — an inference it can get wrong, and says so. A single range with a caveat would have been the easier page to build and the less honest one.
It isn't only typing
The same principle shows up everywhere on this site that an input device is part of the instrument, just with less data behind it because nobody has run a 200,000-person study on it. The reaction time test states plainly that a touchscreen typically reports more slowly than a wired mouse, that scores taken on different devices aren't directly comparable, and that includes your own scores across two devices. The aim trainer is more blunt about it still: “a mouse, a trackpad and a finger are three different instruments,” and its published comparison range is explicitly a mouse range that doesn't extend to the others.
None of those three tests correct for input device, and none of them should. There is no way to measure a visitor's specific mouse, thumb or finger from inside a web page, so a correction would be a guess wearing the shape of a number. Naming the limitation is the honest version of the fix that doesn't exist.
Why this argues for mobile-first, not mobile-also
If touch and keyboard produced the same numbers, building for desktop and adapting the layout down for a small screen would cost nothing. They don't, so that order of operations quietly optimizes the test's design, timing thresholds and reference ranges for the input device that happened to be on the designer's desk, and treats everyone testing on the input device most people actually own as an afterthought. Designing for the smallest screen and the input it implies first is not a cosmetic choice here — it's the only order that doesn't quietly favor one instrument over another before a single visitor shows up.