TL;DR:
- Start with 5-8 interviews per customer segment for initial patterns
- Watch for saturation signals: repeated themes, predictable responses, diminishing new insights
- Use the 3-interview rule: if 3 consecutive interviews yield no new insights, you likely have enough
- Quality beats quantity: 8 deep interviews trump 20 surface conversations
- Different validation goals need different sample sizes: problem validation (5-12), solution validation (8-15), pricing validation (15-25)
Table of contents
- Context and why it matters in 2025
- Step-by-step playbook
- Templates and examples
- Metrics to track
- Common mistakes and how to fix them
- FAQ
- Further reading
- Why CraftUp helps
Context and why it matters in 2025
Most PMs either under-interview (making decisions on 2-3 conversations) or over-interview (talking to 50+ people without clear stopping criteria). Both waste time and resources.
The real question is not "how many interviews to validate" but "how do I know when I have enough signal to make a confident decision?" With AI making product development faster, the pressure to validate quickly while maintaining rigor has never been higher.
Success criteria: You can articulate your customer's problem, predict their likely responses to solutions, and identify clear patterns across different user segments. You stop learning new things that would change your product direction.
Step-by-step playbook
1. Define your validation goal and segment scope
Goal: Get clear on what you are validating and for whom.
Actions:
- Write down your specific hypothesis (problem exists, solution fits, pricing acceptable)
- List your target segments (job titles, company sizes, use cases)
- Decide your confidence threshold (directional insight vs. high confidence decision)
Example: "I believe marketing managers at 50-500 person SaaS companies struggle with attribution reporting and would pay $200/month for a solution that connects ad spend to revenue within 24 hours."
Pitfall: Trying to validate multiple hypotheses in one interview round. Focus on one primary question.
Definition of done: You have a written hypothesis, 1-3 specific segments, and a clear decision you will make with the data.
2. Start with the 5+3 rule
Goal: Establish baseline sample size with built-in expansion logic.
Actions:
- Schedule 5 interviews per segment to start
- After completing these, assess if you are seeing patterns
- Add 3 more interviews if patterns are unclear
- Continue in batches of 3 until saturation
Example: For B2B SaaS validation, start with 5 marketing managers, 5 sales ops people, and 5 revenue operations folks. After 15 total interviews, decide if you need more per segment.
Pitfall: Booking all interviews upfront without reviewing interim findings. You might miss opportunities to adjust your questions.
Definition of done: You have a batched interview schedule with review points built in.
3. Track saturation signals during interviews
Goal: Recognize when additional interviews will not change your conclusions.
Actions:
- After each interview, note new insights vs. confirmations of existing patterns
- Track how many interviews mention the same pain points, solutions, or objections
- Flag when you start predicting interviewee responses accurately
Example: By interview 7, you can predict that marketing managers will mention "attribution is broken" and "reporting takes too long." New interviews confirm but do not add depth.
Pitfall: Confusing demographic diversity with insight diversity. Different people can have identical problems and solutions.
Definition of done: You can predict 80% of what the next interviewee will say about your core questions.
4. Apply the 3-interview rule
Goal: Use a systematic stopping criterion.
Actions:
- After reaching initial sample size, continue interviewing
- Count consecutive interviews that yield no new insights
- Stop when 3 consecutive interviews produce only confirmatory data
- Document what "no new insights" means for your specific validation
Example: Interviews 9, 10, and 11 all mention the same 4 pain points you have heard before, use similar language, and react to your solution concept predictably. Time to stop.
Pitfall: Stopping too early because you are eager to build. Make sure your "no new insights" bar is appropriately high.
Definition of done: Three consecutive interviews with only confirmatory insights and no new themes.
5. Validate your stopping decision
Goal: Confirm you have enough signal for your specific decision.
Actions:
- Review your original hypothesis against interview findings
- Check if you can answer your validation question with confidence
- Identify any remaining uncertainty that more interviews could resolve
- Make the go/no-go decision based on current data
Example: You can confidently say "Marketing managers at mid-market SaaS companies have attribution problems, current solutions do not work for them, and they would pay $150-250/month for a working solution."
Pitfall: Continuing to interview because you enjoy the process rather than because you need more signal.
Definition of done: You can make your product decision (build, pivot, stop) based on interview findings.
Templates and examples
Here is a saturation tracking template you can use during your interview process:
# Interview Saturation Tracker
## Validation Goal
Hypothesis: [Your specific hypothesis]
Decision: [What you will decide based on interviews]
Confidence needed: [Directional/Medium/High]
## Interview Log
| # | Segment | Date | New Insights | Confirmations | Saturation Score |
|---|---------|------|-------------|---------------|------------------|
| 1 | Segment A | 2025-01-15 | 5 new themes | 0 | Low |
| 2 | Segment A | 2025-01-16 | 3 new themes | 2 confirmations | Low |
| 3 | Segment A | 2025-01-17 | 1 new theme | 4 confirmations | Medium |
| 4 | Segment A | 2025-01-18 | 0 new themes | 5 confirmations | High |
## Saturation Signals Checklist
- [ ] Same pain points mentioned in 80%+ of interviews
- [ ] Can predict interviewee responses to key questions
- [ ] No new use cases or workflows discovered in last 3 interviews
- [ ] Solution reactions follow predictable patterns
- [ ] Objections and concerns are repetitive
## Stopping Decision
Current confidence: [Low/Medium/High]
Recommendation: [Continue/Stop/Pivot focus]
Rationale: [Why you are making this decision]
Metrics to track
1. Insight velocity
Formula: New insights per interview / Total insights discovered Instrumentation: Track unique themes, pain points, and solution reactions after each interview Example range: Start at 60-80% new insights per interview, drop to 10-20% at saturation
2. Pattern confirmation rate
Formula: (Interviews confirming existing patterns / Total interviews) × 100 Instrumentation: Count how many interviews reinforce vs. challenge your emerging hypotheses Example range: 20-40% confirmation early, 80-90% at saturation
3. Response predictability score
Formula: (Correctly predicted responses / Total key responses) × 100 Instrumentation: Before each interview, predict responses to your top 3 questions, then score accuracy Example range: 30-50% predictability early, 80-90% at saturation
4. Segment coverage ratio
Formula: Interviews per segment / Target interviews per segment Instrumentation: Track interview distribution across your defined customer segments Example range: Aim for 80-120% coverage per segment (5-8 interviews if targeting 6 per segment)
5. Decision confidence level
Formula: Subjective 1-10 rating of confidence in making your validation decision Instrumentation: Rate your confidence after every 3 interviews Example range: Start at 3-4, reach 7-8 before stopping (8-9 for high-stakes decisions)
6. Time to saturation
Formula: Days from first interview to saturation signal Instrumentation: Track calendar time and interview frequency Example range: 2-4 weeks for focused validation, 4-8 weeks for complex multi-segment validation
Common mistakes and how to fix them
-
Interviewing until you hit a round number (10, 20, 50). Fix: Set sample size based on saturation signals, not arbitrary targets.
-
Stopping after first few interviews confirm your hypothesis. Fix: Actively seek disconfirming evidence and interview skeptical users.
-
Continuing interviews because you found one outlier response. Fix: Distinguish between genuine new insights and statistical noise.
-
Using the same questions across all interviews without iteration. Fix: Refine questions as you learn, but maintain core validation questions for comparison.
-
Interviewing only friendly, accessible users. Fix: Deliberately recruit users who might disagree with your hypothesis.
-
Confusing demographic diversity with insight diversity. Fix: Focus on behavioral and needs-based diversity within your target segments.
-
Stopping because scheduling gets difficult, not because you reached saturation. Fix: Separate logistical challenges from research completeness.
-
Over-interviewing because you enjoy the conversations. Fix: Remember interviews are a means to a decision, not an end in themselves.
FAQ
How many interviews to validate a completely new product idea?
Start with 8-12 interviews across 2-3 customer segments. Focus on problem validation first. You need enough signal to confidently say the problem exists and is worth solving before moving to solution validation.
Should I interview more people if my first interviews are negative?
Yes, but strategically. If early interviews suggest no problem exists, interview 3-5 more people who you think would be most likely to have the problem. If they also say no, you likely have your answer.
How many interviews to validate pricing for a B2B product?
15-25 interviews minimum for pricing validation. Price sensitivity varies more than problem recognition, so you need broader coverage. Include different company sizes and budget holders.
Can I validate with fewer interviews if I use surveys too?
Yes. Use surveys for breadth (100+ responses) and interviews for depth (5-8 per segment). Surveys help validate patterns you discovered in interviews across larger populations.
How many interviews to validate if I am targeting multiple customer segments?
5-8 interviews per segment, not total. If you are targeting marketing managers and sales managers, that means 10-16 interviews total. Each segment needs independent validation.
Further reading
-
Nielsen Norman Group's research on qualitative sample sizes - Foundational research on when qualitative insights reach saturation in user research contexts.
-
Harvard Business Review on customer discovery - Steve Blank's framework for systematic customer development and validation processes.
-
Qualitative research methodology guide - Comprehensive guide to planning, conducting, and analyzing qualitative user research interviews.
Why CraftUp helps
Knowing when to stop interviewing is just one piece of systematic product validation.
- 5-minute daily lessons for busy people cover customer interview questions, problem validation scorecard methods, and JTBD interview questions framework product insights that help you extract maximum value from every conversation
- AI-powered, up-to-date workflows PMs need include templates for tracking saturation, decision frameworks for different validation goals, and scripts that help you avoid validation paralysis start building faster
- Mobile-first, practical exercises to apply immediately let you practice interview analysis, saturation recognition, and validation decision-making with real scenarios
Start free on CraftUp to build a consistent product habit: https://craftuplearn.com
