Most first-time founders fall in love with an idea and then go looking for confirmation. They pitch to friends, family, and anyone who will listen. Everyone nods and says "great idea," and the founder walks away certain. That certainty is usually false. Customer discovery is the discipline of doing the opposite: leaving your desk, talking to real potential users, and structuring the conversation so people cannot flatter you into building the wrong thing.
Why people lie to you, politely
Steve Blank, who defined the Customer Development method, put it plainly: there are no facts inside your building, so get out of the building. But leaving your desk is not enough if you ask leading questions. Ask "Would you use an app that does X?" and almost everyone says yes, because saying no to your face feels rude. This is sharper in the Indian context, where a warm "haan, achha hai" or "sure, I will try it" is often courtesy, not a commitment. Compliments, hypotheticals, and vague future promises are the three sounds of a conversation quietly going nowhere.
The Mom Test
Rob Fitzpatrick's book The Mom Test offers the fix. The name comes from a test for your questions: they should be good enough that even your own mother, who loves you and wants to be encouraging, cannot mislead you, because you never actually mention your idea. Fitzpatrick reduces good discovery to three rules:
- Talk about their life, not your idea. The moment you describe your solution, the other person switches into polite-feedback mode. Keep the spotlight on their problems, habits, and workarounds.
- Ask about specifics in the past, not opinions about the future. "Would you pay for this?" invites a guess. "What did you do the last time this happened, and what did it cost you?" reveals real behaviour.
- Talk less, listen more. If you are talking more than a third of the time, you are pitching, not learning.
Bad questions versus good questions
- Bad: "Do you think this is a good idea?" Good: "Tell me about the last time you faced this problem."
- Bad: "Would you buy a product that did this?" Good: "How are you solving this today, and what do you dislike about it?"
- Bad: "Would you pay 500 rupees a month for it?" Good: "What are you spending on this right now, in money or in time?"
How to run the conversation
Find people who genuinely have the problem, not just friends who want to support you. Reach them wherever they already gather: WhatsApp groups, local trade associations, LinkedIn, college networks, or simply by standing in the market, shop, or office where the problem occurs. Keep it casual and short, around fifteen to twenty minutes. Open with their world: "Walk me through how you currently handle this." Ask "why" and "tell me more" often. Resist the urge to correct them or to reveal your solution until the very end, if at all. Take notes in their exact words, because their language is your future marketing copy.
Look for commitment, not compliments
Praise is data about your ego, not your business. Real signal is when someone gives up something they value: their time for a follow-up, an introduction to their boss, a spot on a waitlist, an advance, or a serious "when can I get this?". If a conversation ends with warm words but no next step, treat it as a soft no. The goal of each chat is not to hear that your idea is good, but to learn whether the problem is real, frequent, and painful enough that people are already spending money or effort to solve it.
Common mistakes to avoid
- Pitching instead of listening, then counting polite nods as validation.
- Talking only to friends and family, who are the worst possible sample.
- Asking about the imaginary future instead of the observable past.
- Ignoring answers that contradict your idea because you have already decided to build it.
Do this for twenty to thirty real conversations before you write serious code. It is the cheapest, fastest way to avoid the single most common startup failure: building something nobody actually wants.