A product thinker wrote this week that founders treat hypothesis validation like a box to tick before they go ahead and build anyway. A few interviews, a quick survey, a landing page, then the magic words: great, we validated it. His correction was simple and brutal. You do not run experiments to prove that you did validation. You run them to learn whether the product is actually needed, and that means being genuinely open to the answer being no.
His proposed fix is the part worth stealing. Instead of trying to prove the idea will work, start by assuming it will not. Then ask what evidence would prove you right, and more importantly, what evidence would prove you wrong.
Why does confirmation-first validation always succeed?
Because every weak signal can be read as encouragement when encouragement is what you are shopping for. An interviewee being polite becomes interest. A survey full of hypothetical yeses becomes demand. Twenty newsletter signups become traction. None of these cost the other person anything, so none of them can carry the weight being put on them, but a founder looking for a green light will see one in any of them.
The inversion changes the evidence bar automatically. If your job is to kill the idea, a compliment does nothing for you. Only costly behavior can save the idea now. Someone bringing you their real task. Someone paying, prepaying, or spending their reputation introducing you to their boss. The things that fail to kill an idea are exactly the things worth building on.
What does trying to kill an idea look like in practice?
Write the obituary first. One paragraph, dated two years from now, explaining why the idea died. A venture investor described this week how she uses the same trick on founder references: instead of asking for weaknesses, she asks them to imagine the company failed and explain what went wrong. People are vague in verdicts and honest in scenarios. The question works on your own idea for the same reason.
Then turn the obituary into tests. If the imagined cause of death is nobody actually pays, the test is a price on a page, not a survey about willingness. If it is the problem is annoying but not urgent, the test is whether strangers who have the problem already built a workaround. No workaround usually means no urgency. If it is we cannot reach these buyers, the test is whether you can get ten of them to answer anything at all.
Is killing an idea a failure?
It is the system working. Going from zero to minus one, building something nobody needed, costs a year and whatever savings funded it. Going from zero to a dead idea in two weeks costs two weeks. The founders who learn this early get more attempts at the real thing, which is the only statistic that matters for a portfolio of one life.
The emotional trap is that validation done honestly threatens the ego, so most founders quietly lower the bar when they like the idea. That is why the inversion has to be structural, decided before the first conversation, not a mood you promise to hold. You are not asking the market whether you are smart. You are asking it to shoot, and checking what survives.
Key takeaways
- Validation that sets out to confirm always confirms. Invert it: assume the idea fails and hunt for the evidence that would prove you wrong.
- Under inversion, compliments stop counting automatically. Only costly behavior can save an idea.
- Write the idea's obituary first, then turn each cause of death into a concrete test.
- A dead idea in two weeks is a win. A dead idea after a year of building is the expensive version of the same answer.
Why this today: a product educator's zero to minus one post and an investor's premortem reference question landed hours apart this morning, making the same argument from two different seats.