How many participants does a study need?

1There is no single right number of participants, because the answer depends on what kind of question a study is trying to answer. 2For finding usability problems, the well-known guidance is about five people per round: if a typical problem trips up roughly a third of users, five sessions give about an 84% chance of seeing it at least once. 3That maths only covers fairly common problems, so rare issues, several distinct user groups or a need for statistics change the picture, which is why card sorts call for about 15 people per group and quantitative studies at least 20. 4Interviews follow a different logic called saturation, where you keep talking to people until new conversations stop changing your main themes, and one widely cited study found its themes had settled by about 12 interviews. 5That is why many teams start with five or six interviews, analyse as they go, and keep recruiting only while new themes are still appearing. 6Whatever the number, who you talk to matters more than how many, because five convenient colleagues tell you far less than five people who really match your audience.