Talk to users before you build anything動手做之前，先跟使用者說話

The most expensive way to learn what people want is to build it first and find out after launch. The three sources below share a cheaper method: go have real conversations, and know how to listen.

Live in the future, then notice what is missing

Paul Graham argues the best startup ideas are noticed, not brainstormed. You put yourself among real users, watch what they struggle with, and the idea appears as a gap in their day. Talking to users is not validation theater; it is where the idea itself comes from.

Ask about their life, not your idea

The Mom Test’s rule is blunt: never pitch, and never ask “would you use this?” People lie to be polite. Ask what they did last week, what it cost them, what they tried before. Facts about the past are evidence; opinions about the future are noise.

Turn conversations into a number

Superhuman’s team asked users one question: how would you feel if you could no longer use the product? Tracking the share who answered “very disappointed” turned fuzzy feedback into a metric they could improve deliberately, segment by segment. Conversations scale when you score them.

What you need before your first ten interviews

• A list of ten people who actually have the problem, with names
• Five questions about their past behavior, zero questions about your idea
• A promise to yourself: no pitching until the interview ends
• One sentence after each call: what surprised you
• A simple score you will track across conversations

The takeaway

• Ideas come from noticing real gaps, so conversations are the source, not the checkpoint
• Ask about past behavior; future opinions are polite lies
• Score your conversations, and fuzzy feedback becomes a metric you can improve