Founders are dangerous interviewers. They know the answer they want and can usually obtain it in twenty minutes.
The customer smiles. The founder writes “strong demand.” Nobody changes what they do.
A compliment is not evidence with good manners. It is still a compliment.
Leave the pitch until the end
Begin with the person’s work, not your product. Ask what happened the last time they faced the problem. Where were they? What started it? What did they use? Who else became involved? How long did it take? What went wrong?
Past behavior has weight. A hypothetical future is almost free.
Avoid questions such as “Would this save you time?” and “Would you pay for this?” They contain the preferred answer. Ask how much time the last attempt took and what they paid for the present solution.
Follow the objects
When someone mentions a spreadsheet, template, inbox, report, or workaround, ask to see it. The object remembers details the speaker omits.
Watch the order of the work. Note handoffs, copied values, checks, delays, and moments where someone asks permission. A product opportunity often lives between the steps people remember to mention.
Keep private information private. Let the user redact or recreate sensitive examples. Observation is not permission to collect everything.
Ask for stories, then numbers
“How often does this happen?” may produce a clean guess. “When did it happen last?” produces a date. Ask for the case before the average.
Useful follow-ups are plain:
- What made that difficult?
- What did you try first?
- Why did that not work?
- What happened because it was late or wrong?
- Who cared enough to complain?
Silence helps. Do not answer your own question because the pause feels long.
Test commitment in the real world
At the end, show the narrow product or describe the test. Ask for a small concrete next step: use it on a real job, introduce the person who owns the problem, provide a safe sample, book a working session, or pay for a pilot.
Interest without movement is weak evidence. A refusal with a reason is useful evidence. “Come back later” needs a date and condition before it belongs in the pipeline.
If you still need people to interview, start with the first ten users guide. If willingness to pay is the unknown, use the early pricing guide rather than turning every research call into a disguised sales pitch.
Write facts, not vibes
After each conversation, record observed facts, direct language, assumptions that changed, and the next commitment. Keep interpretation separate.
After five or ten interviews, look for repeated situations rather than requested features. Different feature requests may point to the same broken job. Your task is to understand the job, not let customers design the interface by committee.
YC’s guidance on how to talk to users emphasizes specifics from the user’s life over hypotheticals. Hacker News discussions return to the same lesson: early interviews work when the founder stops selling long enough to learn what is already true.