How many participants do you need for user research?

How many participants you need for usability testing and user research, by method: qualitative saturation, the 5-user rule, and quantitative sample sizing.

Sample size depends on the research method. Usability testing has a well-known heuristic, qualitative interviews use saturation, and quantitative research depends on the confidence a decision requires.

The usability testing answer: five users, tested in rounds

For moderated usability testing, the most cited starting point is five participants. Nielsen Norman Group's model estimates that five participants from a single user segment will uncover about 85 percent of an interface's usability problems. Additional participants in the same round are increasingly likely to surface problems already found because usability issues are not evenly distributed across users.

The heuristic has two conditions. First, it assumes one relatively homogeneous user segment. If you are testing with two distinct groups, such as new users and power users, plan for roughly five from each group. Second, it assumes iterative testing. NN/g recommends three or four rounds of about five users, with fixes between rounds. This approach lets you test whether each set of changes worked before shipping.

This heuristic is designed to find usability problems. It does not provide a statistically defensible measure of task success rate or time on task. Those questions require a quantitative sample sized for the precision you need.

Qualitative interviews: plan around saturation

Most qualitative research, including discovery interviews, win-loss interviews, and concept feedback, does not have a single agreed-upon sample size. Researchers plan around saturation: the point at which additional interviews stop introducing new themes and mostly confirm existing ones.

Set a target range before fieldwork, then monitor whether new themes continue to appear. A narrow question with a fairly homogeneous audience, such as how existing customers use one feature, tends to reach saturation with a smaller group. A broad question across a diverse audience, such as why prospects choose a competitor, often requires more interviews. Many researchers plan five to eight interviews per segment for a focused question and extend into the teens for broader topics or multiple distinct segments.

Plan coverage per segment. A study spanning three meaningfully different user types has three separate saturation patterns, so a large total can still leave one segment under-covered. Set a target for each segment, then add those targets to determine the overall recruiting goal.

Quantitative research: size for the decision

For quantitative studies using rating scales, NPS, or multiple choice at scale, sample size is based on confidence. Decide how sure you need to be that the result reflects the target population rather than sampling noise. The required confidence depends on how the result will be used.

A directional comparison, such as whether a redesigned onboarding flow feels clearer than the old one, may be useful with a modest sample. An expensive or hard-to-reverse decision requires a properly powered sample calculated against the desired confidence level and margin of error before launch. Ask what happens if the estimate is wrong to determine the level of certainty you need.

Account for the segments you plan to analyze. Splitting a survey with an n of 100 by three plans, two regions, and new versus returning customers can leave some cells with only a handful of responses. Decide which comparisons matter before recruiting, then ensure each relevant cell has enough responses.

What pushes the number up or down

Check these variables before setting a target:

• Number of segments. Each meaningfully different user group needs roughly its own count, whether you are running the five-user heuristic, chasing saturation, or powering a survey cell.
• Novelty versus validation. Testing a brand-new concept tends to surface problems faster, and with fewer participants, than validating something that is already mostly working.
• How reversible the decision is. A UI tweak you can adjust next sprint needs less certainty than a repricing decision or a positioning change.
• Single round versus iterative. Several small rounds let you apply and test fixes between usability or design-validation sessions.
• Moderated versus unmoderated. A moderated study lets you probe in the moment, which is part of why it can work with fewer participants than an unmoderated or purely quantitative approach; see Moderated vs. unmoderated usability testing for the tradeoffs.

Quick reference

Recruiting the number you land on

Once you have a target, choose a recruiting method. In Versive, you can share a study link, embed it in your product, invite contacts from your organization's Audience, or recruit consumers and professionals through integrated Prolific and Respondent providers. See How to recruit research participants (4 ways) or the Recruiting participants docs.

For panel recruiting, screener questions qualify people against the target audience before they reach the study, helping each paid participant count toward the planned segments. A common incentive baseline is about a dollar per minute of study time, with B2B audiences typically requiring more. See Research incentives: what to pay participants for budget guidance.

For a very early concept, AI personas can provide an initial review before recruiting is feasible. Their output does not replace evidence from the participant samples above, but it can help narrow what to test with real people. See What are synthetic users, and when should you trust them? for appropriate uses and limits.