Article · 10 min read

How Many Customer Interviews Are Enough? The Pre-Testing Sweet Spot

Kromatic Free Intermediate

This tackles the how many question head on with a level head. It makes the case that below five conversations you cannot trust anything, ten is a workable minimum, and around thirty removes most of the small-sample penalty. The useful mental shift is to hunt for big obvious signals (like zero of ten confirming your assumption) that let you decide fast rather than chasing statistical certainty.

• Fewer than five interviews is too few to take seriously
• Ten is a workable minimum, roughly thirty removes most small-sample error
• Look for strong lopsided signals so you can act without perfect certainty