How Many Participants Does a Usability Test Need?

What the five-user guidance says

The rule comes from Jakob Nielsen's article "Why You Only Need to Test with 5 Users", published in March 2000 ¹. Nielsen argued that testing with five users finds about 85 percent of the usability problems in a design, and that each additional user adds less and less, because they mostly reveal problems that earlier participants already showed.

The article does not say that a project needs only five participants in total. Its recommendation is to spend the budget on several small rounds. Nielsen recommends splitting a budget into three studies of five users each rather than running one large study. After each round, the team fixes the problems found and tests the revised design, which uncovers problems that were hidden behind the first ones.

The conditions behind the number

The five-user figure rests on assumptions that are easy to forget.

One audience at a time

The guidance assumes participants are reasonably similar. If your site serves distinct groups, such as buyers and sellers, or patients and clinicians, each group may meet different problems. Nielsen's article suggests three to four users from each category when testing two groups, and three users per category when testing three or more groups, so the total grows with the number of groups.

Finding problems, not counting them

Five users can show you that a problem exists and why it happens. They cannot tell you how many of your visitors share it. That difference is the central point of the next section.

Iteration

The 85 percent figure describes one round of one design. The value comes from repeating the cycle: test, fix, test again.

Qualitative and quantitative studies are different

Raluca Budiu of the Nielsen Norman Group addresses the most common misuse of the rule in "Why 5 Participants Are Okay in a Qualitative Study, but Not in a Quantitative One" ². Her point is that the five-user guideline applies to qualitative studies only.

A qualitative study aims to find problems and understand their causes. If one participant out of five struggles to find the pricing page, that is worth investigating, regardless of whether the true rate across all visitors is 10 percent or 40 percent.

A quantitative study aims to measure something with useful precision: a task success rate, average time on task, a satisfaction score. With five participants, the margin of error around any of those numbers is so wide that the result is almost meaningless. Budiu explains that quantitative studies need larger samples, in most cases more than 30 participants, and that confidence intervals only become reasonably narrow with around 40 or more.

The card sorting figure comes from Nielsen's original article, which recommends about 15 users for card sorting and at least 20 for quantitative studies ¹. Budiu's later article sets a higher bar for quantitative work. When planning a measurement study, use a sample size calculation based on the precision you need rather than any rule of thumb.

Planning a qualitative study budget

Suppose you can afford fifteen sessions. Following the guidance above, you would run three rounds of five rather than one round of fifteen. A realistic plan looks like this:

1. Round one: five sessions on the current site. Find and fix the most serious problems.
2. Round two: five sessions on the revised design. Confirm the fixes worked and find the problems the first round hid.
3. Round three: five sessions focused on the areas that changed most, or on a second audience group.

Each round needs time for analysis and for your team to make changes. Our research service plans this cycle around your release schedule.

How to report results from small samples

Small-sample findings are valuable, but they need careful wording. Report counts, not percentages: "three of five participants" is honest; "60 percent of users" invites readers to treat the number as a statistic. Describe what happened and why, and pair each finding with the evidence.

Avoid statements that generalize to all visitors unless you have quantitative data to support them. "Participants who tried to compare plans could not tell them apart" is a solid qualitative finding. "Most visitors cannot compare plans" is a claim the study cannot support.

When five is too few

Even for qualitative work, some situations call for more participants per round:

• The audience is highly varied, such as a public-sector service used by people of many ages, abilities and levels of digital confidence.
• The interface is large and the study covers many tasks, so each participant can only attempt some of them.
• The study includes participants who use assistive technology, and you want to cover several kinds of tools, as described in our accessibility testing guide.
• The stakes are high, such as a medical or financial task where a rare problem could cause real harm.

In those cases, add participants deliberately and explain why in the plan.