I've been talking to a lot of founders lately — people at different stages, different industries, different levels of traction. And the same thing keeps coming up. Everyone wants to know how to validate their idea. How to tell signal from noise early, before they've committed years to the wrong thing.
It's the right question. I just think most people are looking for the answer in the wrong place.
Almost every founder I've spoken to has had someone tell them their idea is great. Almost none of them have had their idea be great in exactly the way they thought.
The problem with early validation is that it tends to come from the wrong people.
Friends, former colleagues, people who respect you — they will be encouraging. That encouragement feels like signal but it's actually noise. What they're validating is your enthusiasm and your relationship with them, not the idea. These are related but very different things.
Paul Graham wrote about something adjacent to this in his essay on schlep blindness — the tendency to unconsciously avoid ideas that involve difficult, unglamorous work. But there's a complementary failure mode I keep seeing in these conversations: clarity blindness. The idea sounds clear because you've been living with it. You've resolved all the internal contradictions in your head. When you describe it to someone who's being kind, their questions don't surface the problems — they give you the benefit of the doubt on the hard parts. So you leave the conversation thinking you've been validated when you've really just been agreed with.
The test that actually works is behavioral. Not "would you use this?" but "here it is — use it." Not "would you pay for this?" but "here's a payment link." The gap between stated intent and actual behavior is where most startup ideas die. People will tell you they have this problem all the time. And then when the product exists and the decision is real, the conversion numbers tell a different story.
This isn't a counsel of despair — it's a counsel of speed. The faster you can get to the behavioral test, the faster you find out whether you have a real problem or a comfortable one. A real problem is one people will change their behavior to solve. A comfortable problem is one that's worth complaining about but not worth the friction of doing something about it.
That pull is real and it's the enemy of early learning. The idea isn't real until someone who has no relationship with you, and no incentive to be kind, changes something about their behavior because of it. Everything before that is a hypothesis.
The founders I've seen navigate this well share one habit: they treat early conversations not as validation exercises but as experiments in confusion. They're not trying to get a yes — they're trying to find where the idea breaks down in someone else's head. The breakdown is the data.
Treat it that way, and the feedback you collect will actually tell you something.