The top starting risks in any business model stem from your customer and problem assumptions. You get those wrong and it’s easy to see how the rest of the canvas falls apart. The best way to validate your customer and problem assumptions is through face-to-face problem interviews.

Surveys aren’t the right tool for problem discovery.

• surveys assume you know the right questions to ask. And because surveys are multiple choice, you also need to know the right possible answers to list on the survey. At the earliest stages of a project, we don’t know what we don’t know…
• surveys can be used for problem validation once you know the right questions and answers to validate, but until then you have to start with a more open ended approach to problem discovery.

It’s all about discovery before validation.

Most of the problems entrepreneurs initially list on their canvas usually don’t turn out to be the right ones. Because most entrepreneurs already have a solution in mind which they can’t simply shut off. Instead of asking: “What are my customer’s top problems?”, many entrepreneurs ask: “What top problems can I solve with my solution?”

When you have already decided to build a hammer, everything starts looking like a nail, and you fake the problems on the canvas to justify your solution.…When you put those same problems on a survey and ask people to rank them, sure they can rank them relative to the other choices on your survey. But if their top problem isn’t there, they have no way of letting you know, and you never discover it.

Even if you get people responding that they have a problem, you don’t get to the real “why” on a survey. The real “why” is often several levels deep and getting to it requires a conversation. You don’t know what they have tried so far, why it didn’t work, etc. Knowing these details is key to building a mafia offer later.

At early stage, finding evidence of monetizable pain is your number one priority and running one-on-one problem interviews is your best course of action. One-on-one interviews may not seem efficient, but they pack more learning per unit time than anything else you could possibly be doing. You also don’t need as many data points as you might think to start finding actionable patterns.

When you can start predicting what people are going to say before they say it, that’s when you know you’re done. It usually takes a minimum of 10 interviews and sometimes up to 30 interviews to get there.