Skip to main content
Evidnt

How Many Interviews Do You Need?

There's no magic number, but there are clear guidelines for knowing when you have enough data — and when you need to keep going.

Last updated 2026-02-24

The short answer

Start with 5. Aim for the number Evidnt recommends. Stop when patterns stabilize.

During the context analysis step, Evidnt recommends a minimum number of interviews for your specific situation — anywhere from 5 to 20. This recommendation is calibrated to your market:

  • Niche B2B (e.g., CTOs at fintech companies) → fewer interviews needed (5-8), because your market is small and patterns emerge faster
  • Mass-market B2C (e.g., college students who cook) → more interviews needed (15-20), because your audience is diverse and you need broader signal

The 5-interview rule

After 5 conversations, you'll start seeing themes. Not conclusions — themes. Someone's frustration with manual invoicing sounds a lot like the person before them. A pricing objection echoes across three interviews.

This is when validating an idea starts getting interesting. The first 2-3 interviews feel scattered. By interview 5, the picture sharpens.

But 5 interviews don't prove anything. They show you where to look harder.

Quality over quantity

5 deep interviews with tagged insights beat 15 shallow conversations with no structure.

A "deep" interview means:

  • You asked follow-up questions when something interesting came up
  • You captured detailed notes (at least 50+ characters for extraction by Evidnt)
  • You tagged multiple insights across different categories
  • The interviewee fits your ideal profile (not a random contact)

Evidnt measures interview quality with a specific threshold: each interview should have 2 or more insights across 2 or more categories. This ensures your conversations are producing structured, analyzable data.

The validation gate

To get a validation score (Pro), you need to meet two conditions:

  1. Interview count — Complete at least the Evidnt-recommended minimum (5-20 interviews)
  2. Interview quality — Every completed interview has 2+ insights tagged in 2+ distinct categories

Both conditions must be met. You can track your progress on the Patterns page, where a gate checklist shows which requirements you've met and what's remaining.

Signs you've done enough

  • Patterns are repeating. New interviews confirm what you've already heard rather than revealing new themes.
  • Validation score has stabilized. Your score isn't swinging dramatically with each new interview.
  • You can predict responses. Before the interviewee answers, you have a reasonable guess about what they'll say.
  • Category balance exists. You have insights across multiple categories, not just pain points.

Signs you need more interviews

  • Patterns keep shifting. Each new interview changes the picture significantly.
  • Conflicting signals. Half your interviewees love the problem space, half don't recognize it.
  • Category gaps. You have lots of pain point data but almost no pricing or objection signals.
  • Low or unstable validation score. A score in the "Explore" range (31-55) often means you need more data, not that the idea is bad.
  • Small sample from one segment. If your target customer has sub-segments, make sure you've covered the key ones.

What about diminishing returns?

There's a point where additional interviews add less new information. In practice:

  • Interviews 1-5: Rapid learning. Every conversation reveals something new.
  • Interviews 6-10: Pattern formation. You start connecting dots.
  • Interviews 11-15: Pattern confirmation. New data reinforces or challenges existing themes.
  • Interviews 15+: Diminishing returns for most markets — unless you're in a very diverse B2C space.

Evidnt's validation score naturally reflects this. Once your score stabilizes across multiple interviews, you've likely reached the point of sufficient evidence.

If you haven't started your interviews yet, describe your idea on Evidnt — you'll get your questions and outreach kit in minutes. Free, no account needed.

Next step

Once you have enough data, the next question is what to do with it: When to Pivot vs. Persevere.

Related