Customer Interviews Can Mislead You: Test Behavior Before You Build
Customer interviews are useful, but polite interest is not demand. Test behavior, commitment, and willingness to pay before treating a conversation as validation.
Why interviews are useful for learning about problems, but behavior is stronger evidence of demand.
In short
- • Interviews reveal problems and language but do not automatically prove demand.
- • Existing workarounds and commitments provide stronger evidence than hypothetical enthusiasm.
- • Use progressively stronger behavioral tests to reduce uncertainty.
Customer interviews are valuable, but an interview can prove that a person understands a problem without proving that they will buy a solution.
The practical mistake is treating stated enthusiasm as demand.
Start with the last real occurrence
Instead of asking:
"Would you use this?"
Ask:
- When did this problem last happen?
- What did you do?
- What did it cost?
- What workaround did you use?
- Who was involved?
- What have you already tried?
- What happened afterward?
Real events provide context that hypothetical answers often lack.
Separate problem evidence from solution evidence
A customer can have a serious problem and still reject your proposed solution.
Track the distinction:
Problem evidence: the problem exists and matters.
Behavior evidence: the customer already acts to manage it.
Solution evidence: the proposed approach changes the situation.
Commercial evidence: the customer is willing to commit resources.
These are different claims and require different evidence.
Use an evidence ladder
A useful ladder is:
- Problem recognition.
- Existing workaround.
- Cost or consequence.
- Access to the real workflow.
- Concrete commitment.
- Actual usage.
- Economic outcome.
The strongest signal depends on the business model. A paid pilot can be powerful for a B2B service. A repeated usage pattern may matter more for a consumer product.
Do not over-read interview language
Phrases such as:
"I would definitely use this."
"That sounds great."
"We need something like this."
are useful signals.
They are not equivalent to:
"I will give you access to the workflow."
"I will introduce the decision maker."
"I will pay for the pilot."
"I will use this with real work."
YC's startup guidance recommends direct customer contact and iteration rather than relying on a perfect pre-launch plan.
Test the smallest solution
After learning from interviews, run a behavioral test.
Possible tests:
- manual service;
- prototype;
- paid pilot;
- deposit;
- real-data workflow;
- scheduled implementation;
- limited trial.
The test should answer the most important uncertainty.
If the uncertainty is willingness to pay, a usability test is not enough.
If the uncertainty is whether the workflow can be delivered, a survey is not enough.
Community evidence
Founder discussions often describe conflicting signals from interviews, forums, and early customers. These discussions are useful for discovering questions but are not representative surveys.
Use them to identify failure modes:
- polite agreement;
- leading questions;
- founder pitching too early;
- asking hypothetical purchase questions;
- ignoring existing behavior;
- confusing a feature request with demand.
Founder experience should remain bounded
If you use a founder story, record the context.
One founder's experience may depend on:
- customer type;
- market;
- price;
- timing;
- distribution;
- founder reputation;
- business model.
Do not turn a successful interview tactic into a universal rule.
What commonly goes wrong
Do not:
- pitch during every interview;
- ask only positive questions;
- collect compliments as validation;
- ignore people who disagree;
- treat Reddit discussion as market research;
- ask for future behavior without checking past behavior;
- build a large product before testing the central assumption.
Decision framework
Before building, write:
Hypothesis: what must be true?
Evidence: what have we observed?
Gap: what is still uncertain?
Test: what behavior would reduce the uncertainty?
Threshold: what result changes the decision?
Next move: build, change, test again, or stop.
Related SOE Knowledge
Connect this with SOE's demand validation, first customer, feature prioritization, and product-market fit material.
Sources and methodology
Y Combinator's published startup guidance provides the primary evidence for customer contact and iteration. Public founder discussions are qualitative discovery material only. The evidence ladder is SOE synthesis and is not a universal validation score.
Final answer
Interviews are for learning about the problem and the buyer's context. They become stronger evidence when they lead to observable behavior.
Listen to what people say, then test what they do. Build the smallest experiment that can answer the uncertainty that matters most.
Related SOE knowledge
Take it into your work
Do something with what you just read.
Save it for later, mark the passage that matters, or open your SOE workspace and keep thinking.
Takeaways
What to carry forward
- → Ask about recent behavior
- → not hypothetical preferences.
- → Look for existing workarounds and costs.
- → Test meaningful commitments before building extensively.
- → Treat Reddit as qualitative signal
- → not representative proof.
- → Define validation as a testable hypothesis.
Listen carefully to what customers say, then test the behavior that matters.