A Speedup Users Notice Is About 20% of the Wait
Cutting a second from 10 seconds is invisible, while cutting a second from 2 seconds feels transformative.
producttribe.substack.com · The 23 Minutes That Never Were: On Asynchronous Communication, the Productivity Myth, and What Research Actually Shows | Dear UX Designer, the Tool is not the ThoughtTwo engineering teams each cut a second off their product's loading time. One goes from 10 seconds to 9 seconds. The other goes from 2 seconds to 1 second. The effort is identical, and the boss praises both equally, but users experience them completely differently. The first change is barely noticed. The second feels transformative.
The reason is a law of perception discovered in the 19th century. In the 1830s, Ernst Weber, a German physiologist studying weights and sounds, found that the smallest difference people could reliably detect was not a fixed amount but a fixed proportion, about 2% of what they started with. If they could just tell a 102 gram weight from a 100 gram weight, they needed 204 grams to tell the difference from 200 grams.
Decades later, Gustav Fechner, a physicist and philosopher, turned that observation into math. Sensation, he argued, grows with the logarithm of the stimulus, not in a straight line. In plain terms, our senses compress big numbers and amplify proportions. The jump from 1 to 2 feels like the jump from 100 to 200, because each is a doubling.
For websites and apps, SitePoint and other web performance specialists use roughly a fifth of the starting time as a rule of thumb for the smallest change people reliably notice. It is a practical guideline, not an established law. A page that takes 5 seconds must get down to 4 seconds or quicker before people consistently sense it. A 200 millisecond saving is invisible on a 5 second load, but substantial on a half second load.
That creates a simple test. A cut from 4 seconds to 3.5 seconds is a 12.5% improvement, real but likely unfelt. A cut from 4 seconds to 3 seconds is 25%, and likely to register. Small gains can still compound and help the machinery, but they should not be sold as a better experience.
A second off a ten-second wait is easy to miss. A second off a two-second wait changes how the whole product feels.
Questions people ask
Does the 20% rule apply to price and features too?
Roughly, yes. People judge many things proportionally, so a $10 hike stings far more on a $20 item than on a $200 item. The exact threshold varies by context, so 20% is best treated as a first filter, not a universal constant.
Why would our brains work this way?
One likely reason is that a logarithmic sense lets us operate across enormous ranges, from whispers to jet engines and dim starlight to bright sun. It keeps us sensitive to what changed relative to the current background, which is often what matters for decisions.
If a fix is smaller than 20%, should a team skip it?
Not necessarily. Small fixes can add up to cross the threshold together, and they can cut costs, reduce errors, or prevent future slowdowns. The point is to avoid promising users a noticeable difference until the total change is large enough to feel.