How many people do you think you have to speak with to recognize patterns?

In usability, how many people do you think you have to speak with to recognize patterns? Hundreds? Thousands?

Nope. Maybe just four or five.

If, for example, you’re looking to understand how easy or difficult it is for people to book flights on an updated website, 80% of the problems that can be identified will likely be identified with the first four or five people you study. If you don’t identify such problems after five people, you're either asking the wrong question or you're asking the wrong people for feedback.

This assumes, of course, that you have properly defined the segment(s) to gather feedback from. Do you want to talk to people who book 10+ flights a year? Do you want to talk to people who book for entire families? This four or five person rule applies to groups of people with different behaviors, or “usability segments."

Clients are often surprised when I say extraneous participants are a waste of time and resources. Instead of adding more than five participants to any one behavioral segment, my suggestion is to repurpose your budget and invest in either additional segments or additional studies.