Glossary

How many customer interviews do you need?

For qualitative discovery, most studies begin repeating themselves between 12 and 20 interviews within a single well-defined segment.

The number that matters is the count per segment, not the total across all of them.

This is a practical heuristic rather than a statistical rule. Qualitative research does not produce statistically representative findings at any sample size. It produces an inventory of what exists in the population. The stopping condition is thematic saturation: new interviews stop producing new themes.

Why it matters

Teams get this wrong in both directions. Some stop at three conversations and treat one person's workflow as a market truth. Others delay acting until they have 60 interviews spread across five segments, which is 12 per segment, the same evidence they'd have had months earlier with a narrower scope.

Example

Twenty interviews across four segments is five per segment: too thin to distinguish a pattern from a personality in any of them. Twenty in one segment will usually saturate. Narrowing the segment is almost always the better move.

In UpSight

Theme pages in UpSight count distinct people, not quotes, so one person repeating a point five times reads as one. Filter the themes view by a group (a saved rule over your people, such as a role or a company size) and the counts re-scope to that group. That is the per-segment number this page is about.

Related terms

FAQ

Is 5 customer interviews enough?
Five is enough to find obvious usability problems in a defined task. It is not enough to establish that a market problem exists.
How many interviews for statistical significance?
None. Qualitative interviews are not a statistical instrument. If you need significance you need a survey with a designed sample.
When should you stop interviewing?
When two or three consecutive interviews in the same segment produce no new themes.