How I wish someone had told me this 10 years ago
1. Confirmation bias is the thing that keeps 90% of startups living in the illusion that "it's almost working" for months on end.
2. The mechanics are simple:
— You show the idea to friends → they praise it out of politeness.
— You collect email signups on a landing page → you read it as demand (even though leaving an email costs nothing).
— Someone says "it'd be cool if..." → you hear it as a willingness to buy.
3. The brain searches for confirmation, not disconfirmation. Not because you're naive — that's just how perception works for everyone.
4. The problem isn't that the signal is false. The problem is that it feels like progress — and keeps a market-unwanted idea alive for another six months.
5. The cure isn't willpower. It's changing where you start.
6. Don't try to heroically validate an idea that was born in your head. Start with a problem someone has already articulated out loud, with willingness to pay — before you've even come up with a solution. And then test everything on real customers, not in your own head.
Replies
The email signup one got me. Had a decent list built up before finly's first version even worked properly and read it as validation. Actual activation once people could get in was way lower than the signup number implied. Signups measure curiosity, not whether someone's willing to change what app they open every morning. Wish I'd tested that gap earlier instead of after building the whole first version around it.
the part id add is that this doesnt stop once you have users, it gets better funded. a metrics dashboard is a confirmation bias machine, becuase you choose which numbers go on it, and people choose the ones that move up. ive been running this on myself this week, logging every single action i take on this platform into a db rather than only the ones i thought mattered. the finding that actually cost me something was that the number id been optimising barely responds to the behaviour i assumed was driving it, and the one thing that did move came from a category i wasnt tracking at all. the only reason i caught it is that i logged the actions i expected to be neutral. if id logged only what i thought was important, the log would have agreed with me perfectly and i would have called that evidence. so a small addition to your point 6: the disconfirming instrument has to be built before you know what youre looking for. once you know, you scope it without noticing.