“Would you use this?” invites a pleasant answer and little evidence. “How did you handle this the last time it happened?” can reveal an existing workaround, its cost, and the reason a founder's proposed product might not matter.
Customer discovery should help you understand current behaviour before you ask people to evaluate a future feature list.
What is a current-workaround question?
It asks about a specific past situation and the steps someone actually took. The useful detail is what happened, who did the work, which tools were involved, and what remained unresolved.
For example, if you are considering appointment software, ask an office manager to describe the last booking mistake they had to repair. Avoid starting with your solution or assuming that the mistake happens frequently.
Use questions that invite a concrete account
| Question | What it can reveal |
|---|---|
| When did this last happen? | Recency and a real example |
| What did you do first? | The existing workflow |
| Which tools or people were involved? | Dependencies and switching costs |
| What was difficult about that? | Friction in the participant's terms |
| What have you already tried? | Alternatives and past decisions |
| What happened afterward? | Consequences and unresolved work |
Follow the answer instead of reading the entire list mechanically. If a person says the workaround is good enough, investigate why before explaining how your proposed feature would be better.
Keep observations separate from your interpretation
Imagine a fictional respondent says, “We put missed calls in a spreadsheet and check it after lunch.” That establishes a reported process. It does not establish that they want an AI receptionist, have purchasing authority, or would pay your proposed price.
Write notes in separate fields: participant statement, concrete example, observed limitation, founder interpretation, and follow-up question. Keep the original response available so a summary does not become stronger evidence than the participant supplied.
Ask about spending carefully. Existing expenditure can reveal the seriousness of a problem, but hypothetical willingness to pay is not a completed purchase.
Choose a collection format that fits the question
Live interviews allow immediate follow-up and observation. Asynchronous voice or text responses can be easier to collect across schedules, but vague answers may require another contact.
In StackFlight, I built a workflow for sharing studies and reviewing original voice or text responses. The product keeps responses available for the founder's own review; it is not evidence that every study validates a business idea.
First Round's customer-discovery interview offers additional practical context. Use external frameworks to improve your questions, not to replace attention to the participant.
Discovery questions
How many interviews prove an idea will work?
No fixed interview count proves future demand. Look for relevant repeated behaviour and test the next assumption.
Should I show a prototype immediately?
Understand the existing situation first, then clearly separate prototype feedback.
What should I build after discovery?
Choose a narrow workflow supported by evidence. Turn it into a developer brief before commissioning the build.