A customer discovery evidence checklist for founders
Before deciding that an interview validates a problem, separate what the person experienced, what they did about it and what remains your interpretation. A useful conversation can disprove a promising idea. It does not have to end with a sale.
Use this checklist to prepare a call and write up what you learned. It is a practical aid, not a scoring system that proves product-market fit.
Start with the decision the call should inform
Write one sentence: “After this conversation, I want to know whether…” Finish it with a specific uncertainty about a particular type of customer. “Understand the market” is too broad to tell you which follow-up matters.
For example: “I want to know whether small agency owners lose billable time when client requests arrive outside their agreed process.” This is an invented research example, not a claim about agencies.
Four things to establish
| Evidence to look for | Suggested question | What does not establish it |
|---|---|---|
| A real situation | “Tell me about the last time this happened.” | Agreement that the problem could happen |
| Its consequence or lack of consequence | “What happened as a result?” | The founder saying it sounds expensive |
| Existing behaviour | “What did you do about it?” | A hypothetical feature wish list |
| Relative priority | “Have you considered changing this? Why or why not?” | Enthusiasm during a demonstration |
For each row, save a short factual note, the source and the unanswered question. Use participant codes and avoid unnecessary identifying details. Check permission before retaining sensitive quotations. An explicit “it caused no material problem” is an answer. Missing evidence is not evidence that the problem is severe.
Keep fact and interpretation in different columns
Imagine a participant says: “I check two systems every morning because the totals sometimes differ.”
Fact: They report checking two systems each morning.
Interpretation: The checks may be a burden, or an acceptable control they want to keep.
Next question: “What happens when you find a difference?”
Jumping directly to “they need a single dashboard” would hide both the cause and the value of the current process. Ask before assigning a solution.
Record contradictions rather than smoothing them away
Someone may describe a workflow as painful and later say replacing it would create more work. Preserve both statements. Ask how those fit together. The person may prefer a familiar inconvenience, lack authority to change it or have a reason your solution would miss.
A useful check is: “Earlier you mentioned the manual checks take time, and you also said you'd keep them. What makes them worth keeping?” This offers an interpretation for correction rather than arguing with the participant.
Distinguish a proposed next step from an agreement
“Would you try a prototype?” is a question. “I could try it sometime” is tentative interest. A concrete agreement might specify what will happen, who will do it and when. Record only the details actually agreed; do not fill in a date or owner on the customer's behalf.
A research conversation does not need a buying commitment. An appropriate next step may be another interview, an introduction, permission to observe a workflow or no further action.
Reusable interview record
- Customer type and why this person fits:
- Decision this interview informs:
- Recent example and its source:
- Consequence, including evidence of no significant impact:
- Current workaround and previous attempts:
- Competing priorities:
- Contradictions or counterevidence:
- Important unknown and next question:
- Next step: proposed or agreed; owner and timing only if known:
Compare records from similar customers. Five positive comments from unrelated customer types are not one repeatable pattern. Look for repeated circumstances and behaviour, and track the people who did not experience the problem too.
Using the checklist during a live call
You do not need to display the whole worksheet while someone is talking. Keep the objective and the next unresolved question visible, then update the record afterwards. Our follow-up question guide gives examples of moving from a vague answer to something specific.
Truffle's invite-only Mac beta suggests questions during discovery and early sales calls. Coaching requires a separate invitation; the initial setup is an Apple Silicon Mac on macOS 14.2 or later, using Google Meet in Chrome. See an illustrative conversation. Beta setup requirements and limitations are on the installation page; this worksheet is useful without installing anything.
Further reading
The Mom Test is a practical introduction to learning from customer conversations. Huthwaite's SPIN methodology provides a sales-question framework. This checklist is Truffle's own synthesis and should not be treated as either author's named framework.