Discovery questions for founders: what to ask and what not to ask, according to Y Combinator

Good discovery means asking about the customer's problem and how they handle it today, and not mentioning your idea until the end, if at all. In his Startup School talk How to talk to users, Y Combinator group partner Gustaf Alströmer says the most important thing is not to introduce your idea too early, because it can bias the answers: "your role in this interview is to listen, not to talk". The same logic applies in a sales call before a demo.

Which six questions does he suggest?

He gives six questions, where X is the specific task or problem you are trying to solve:

• Tell me how you do X today.
• What is the hardest thing about doing X?
• Why is it hard?
• How often do you have to do X?
• Why is it important for your company to do X? (He says to dive deep here: the motivation is critical.)
• What do you do to solve this problem for yourself today?

For sales conversations, YC's Tom Blomfield adds two in The Sales Playbook For Founders: "What's the part of your job you hate the most?" and, to map the buying process, "Describe the last time you bought software like this. Who internally had to approve it?" The discovery questions tool helps you adapt these to your own market.

How do you keep the person talking?

Use open follow-ups. Alströmer suggests "Tell me about that", "What do you mean by that?", "Can you tell me more about that?" and "Why is that important to you?". If the person can show you how they do the task on their screen, even better: he wants you to understand behaviour, not only what is said, because changing behaviour is hard. And take notes during the call: you will have to turn the conversation into notes anyway.

Which questions should you avoid?

Alströmer lists these as ways to derail the conversation: "Will you use our product?" (people will probably say yes, which means nothing), "Which features would make it better?", yes/no questions, "How would a better version look to you?" and asking two questions at once. His point is that users generally have good problems but bad solutions, so your job is to understand the problem, not to collect feature ideas.

Why do this before the demo?

Because the answers decide what to show and what the product is worth to this person. A copilot that uses your confirmed product context can suggest discovery questions and the next question during the call, so you spend your attention on listening. After the call, the post-call report keeps the customer's own words about the problem. For related mistakes, read 5 common B2B sales mistakes. For the book-length version of this skill, Rob Fitzpatrick's The Mom Test is about getting reliable, unbiased feedback from customer conversations.

Sources: Gustaf Alströmer, How to talk to users (Y Combinator, video); Tom Blomfield, The Sales Playbook For Founders (video); Rob Fitzpatrick, The Mom Test.