What Real Validation Looks Like

You have a hypothesis. Now you need to test it. But most first-time founders test it wrong – they ask leading questions, talk to friends, and count enthusiasm as signal.

Real validation is designed to surface reasons your idea is wrong, not reasons it's right.

The Mom Test

The classic mistake is what Rob Fitzpatrick called "the mom test" problem: asking questions your mom would answer encouragingly even if she'd never use your product.

"Would you use an app that helped you manage your bakery orders?" → "Oh, that sounds wonderful, honey."

That tells you nothing. The question is designed to get a yes. A better question looks like this:

"Walk me through how you handle a custom cake order right now, from the moment someone calls in to the moment it's picked up."

That question can't be answered encouragingly. It produces a description of reality – and reality is where your hypothesis either holds or breaks.

What You're Looking For in a Conversation

A good validation interview has one job: find out if the problem you think exists is the problem they actually have.

Strong signal	Weak signal
They describe the problem without being prompted	They agree when you describe it to them
They've tried to solve it already (workaround exists)	They say "I just deal with it"
They can name a specific moment when it cost them	They say "yeah it's a bit annoying"
They ask when they can sign up	They say "I'd probably try it"
They refer you to someone else with the same problem	They say "I know a few people who might need this"

The strongest signal of all is the workaround. If someone built a spreadsheet, hired a part-time person, or uses three disconnected apps to solve the problem you're describing – the problem is real and painful enough that they took action. That's your early adopter.

The Three Tiers of Validation Evidence

Not all evidence is equal. Here's how to think about what you're collecting:

• Tier 1 – behavioral evidence (strongest): they already spend money or time on this problem. They have a workaround. They've complained about it publicly. They switched tools trying to solve it;

• Tier 2 – conversational evidence (useful): they describe the problem in their own words, unprompted. They get specific about when it happens and what it costs them. They ask questions about your solution;

• Tier 3 – attitudinal evidence (weakest): they say they would use it. They say it sounds useful. They say they know people with the problem.

Maya needs Tier 1 and Tier 2 evidence before she builds anything. Tier 3 is noise.

How Many Conversations Do You Need

The standard advice is 10–20 interviews before you start drawing conclusions. But quality matters more than quantity. Five conversations where someone describes a real workaround and asks how to sign up is stronger evidence than twenty conversations where people say "yeah, that sounds useful."

The signal you're looking for is convergence: when multiple people, without prompting, describe the same problem in the same way and point to the same moment it breaks. When that happens three times, you have something.