Customer Discovery and Validation in 2026
How does an entrepreneur do customer discovery and validation in 2026? Customer discovery is a structured process of talking to 20-50 potential buyers, asking about past behavior (not future opinions), and validating willingness to pay before building anything. AI tools can accelerate research and synthesis, but live conversations remain the only reliable signal.
Key Takeaways
- Talk to real humans before writing code or buying ads.
- Ask about past behavior; ignore future intentions.
- Pre-sales (paid commitments) are the strongest validation signal.
- 20-30 quality conversations is the minimum for confident decisions.
- AI helps with synthesis, not the conversations themselves.
The Discovery Funnel
| Scène | Goal | Method |
|---|---|---|
| Problem discovery | Confirm a real pain exists | Open-ended interviews (20-30) |
| Solution discovery | Test what people would buy | Concept tests + landing pages |
| Willingness to pay | Confirm someone will pay | Pre-sales, deposits, LOIs |
| Product validation | Confirm they use what you built | Beta users + retention metrics |
| Channel validation | Confirm scalable acquisition | Paid + organic small tests |
How to Run a Customer Interview
Find 20-30 Real Targets
LinkedIn, niche communities, and warm intros work best. Cold email yields 5-10% response rates with the right pitch.
Ask About Past Behavior
Use questions like “Walk me through the last time you tried to solve X.” Avoid “would you use…” — answers are systematically wrong.
Listen for Emotion
Real pain shows up as frustration, swearing, or storytelling. Polite agreement is meaningless.
Synthesize Patterns
After 10 interviews, themes emerge. Use AI to cluster transcripts and surface common phrases.
Validation Beyond Interviews
- Smoke-test landing page: measure email signups against ad spend.
- Pre-orders: people pay deposit for product not yet built.
- Letter of intent: B2B written commitment to purchase.
- Concierge MVP: deliver the outcome manually before automating.
- Wizard-of-Oz MVP: appears automated but humans run it behind the scenes.
Common Mistakes to Avoid
- Asking “would you use this?” Hypothetical answers don’t predict behavior.
- Talking only to friends. They lie to be supportive.
- Skipping payment validation. Interest doesn’t equal money.
- Building too much before testing. A landing page can validate in days.
Action Steps
- Pick 30 target customers and reach out for 25-minute calls.
- Write 5 past-behavior questions for the conversation.
- Build a smoke-test landing page in one day.
- Ask 10 prospects to pre-order or pay a deposit.
- Synthesize learnings weekly with AI summarization.
FAQ
How many interviews are enough?
20 is a minimum to spot patterns; 30-50 gives confidence. Stop when you stop hearing new things.
Do customer interviews work for B2C?
Yes — but supplement with quantitative tests (ads, surveys, smoke tests). B2C decisions are noisier.
Can I use AI to do interviews for me?
Not yet. AI can transcribe and synthesize, but live human conversation is still required for nuance.
What’s a good pre-sale conversion rate?
For warm leads, 10-25% is healthy. For cold traffic, 1-3%.
What if no one wants to talk to me?
Your value proposition is unclear. Rewrite the outreach pitch in plain language and try again.
Sources & Further Reading
- “The Mom Test” by Rob Fitzpatrick — best book on customer interviews.
- Steve Blank’s Customer Development methodology.
- YC’s Startup School on early validation.
About Riman Agency: We help founders run validation sprints that prevent expensive mistakes. Book a discovery sprint.
← Previous: Funding and Pricing | Series Index | Next: Email and the Owned Audience →