SUS sample size and confidence intervals: how sure is your score?
Updated 2026-11-16
A SUS score from a study is an estimate. If you tested a different group of people, you would get a slightly different number. A confidence interval shows how far the true score could reasonably be from your estimate.
How the 95% confidence interval is calculated
Take the mean of the participants' scores and the standard deviation (how spread out they are). The margin of error is:
margin = t × standard deviation ÷ √n
where n is the number of participants and t is a value from the t distribution for n - 1 degrees of freedom (2.26 for 10 participants, 2.09 for 20, close to 1.96 for large samples). The interval is the mean minus the margin to the mean plus the margin. Our batch calculator does this for you.
How sample size changes precision
The table shows the margin of error for two typical spreads of scores. The standard deviations are examples for illustration; use your own data for real numbers.
| Participants | Margin if SD is 15 | Margin if SD is 20 |
|---|---|---|
| 5 | ±18.6 | ±24.8 |
| 10 | ±10.7 | ±14.3 |
| 15 | ±8.3 | ±11.1 |
| 20 | ±7.0 | ±9.4 |
| 30 | ±5.6 | ±7.5 |
| 50 | ±4.3 | ±5.7 |
With 5 participants, a mean of 70 could easily be anything from the low 50s to the high 80s: that is the difference between an F and an A. With 20 to 30 participants the interval narrows to roughly a grade either side.
What this means in practice
- Formative tests with 5 to 8 people are great for finding problems, but their SUS mean is a rough indication. Report it with the interval, or not at all.
- Benchmarks and before-and-after comparisons need more people. Aim for the interval to be narrow enough to answer your question, for example "are we above 68?".
- Comparing two designs: if the two intervals do not overlap, the difference is very likely real. If they overlap, the difference may still be real; use a proper two-sample test before concluding there is none.
Why a confidence interval helps with stakeholders
Saying "the score is 72, and we are 95% confident it is between 66 and 78" stops people over-reading a two-point change between rounds. It also makes a clear case for testing with more people when a decision depends on the number.
Questions
How many participants do I need for SUS?
There is no single number. Work back from the precision you need: with typical spreads, around 20 to 30 participants give a margin of error of roughly 6 to 9 points.
Can the confidence interval go above 100?
Mathematically the formula can, with small samples, but SUS cannot. Our calculator caps the interval at 0 and 100.