Sample size calculator for UX research
Work out how many participants you need to estimate a score precisely, to detect a difference between two designs, or to find most usability problems.
Free and private: everything is calculated in your browser and nothing you paste is sent anywhere.
How it is calculated
- Estimating a score: n = (t × SD ÷ margin)², recalculated until n settles, because the t value depends on n.
- Comparing two designs: n = k × ((t for confidence + t for power) × SD ÷ difference)², with k = 2 for different people per group and k = 1 for the same people (within-subjects). Again iterated on t.
- Finding problems: the chance of seeing a problem at least once with n users is 1 − (1 − p)ⁿ, so n = ln(1 − goal) ÷ ln(1 − p).
Frequently asked questions
How many users do I need for a usability test?
It depends on the goal. To find problems that affect about a third of users, five users give roughly an 85% chance of seeing each one, which is where the famous "five users" figure comes from. Problems that affect fewer people need more users. To report a score such as SUS with a margin of ±5 points at 95% confidence you need around 50 people, and comparing two designs usually needs more.
How many participants do I need for a SUS study?
With the typical SUS standard deviation of about 17–18 points, roughly 50 participants give a 95% confidence interval of ±5 points, and about 15 give ±10 points. Use the "Estimate a score" mode above with your own margin.
What standard deviation should I use?
The best source is your own earlier data on the same measure. Without that, published values are a reasonable starting point: SUS scores usually have a standard deviation of about 17–18 points. A larger standard deviation always means you need more participants.
What is statistical power?
Power is the chance that a study detects a real difference of the size you specified. 80% is the usual convention (Cohen, 1988), so one study in five would miss a true effect of exactly that size. Raising power to 90% needs about a third more participants.