How to validate a startup idea before you build anything
Most startup ideas do not fail in the market. They fail in the founder's head months earlier, sustained by polite encouragement from friends and a few likes on a launch post. Validation exists to move the failure earlier, when it costs a weekend instead of a year.
The trap is that humans are terrible interview subjects. They compliment, they speculate, they say "I would totally use that" and mean nothing by it. Good validation is designed so that politeness cannot contaminate the data.
Talk to customers without lying to yourself
Rob Fitzpatrick's 2013 book The Mom Test remains the shortest useful manual here. Its core rule: ask about past behavior, not future intentions. "Would you pay for this?" invites fiction. "When did you last hit this problem, and what did you do about it?" invites facts. If the honest answer is "I googled for ten minutes and gave up," the pain is not a business.
Thirty conversations with people who actually own the problem beats three hundred survey responses from people who do not. Recruit them where the pain lives: industry Slack groups, niche subreddits, trade fair hallways, the comment sections of the tools they currently hate. Founders who skip this and interview only other founders get validation for a product aimed at founders, which is rarely the market.
The smoke test: a landing page and a price
Interviews measure pain; a smoke test measures commitment. Put up a page that describes the product as if it exists, attach a real price and a "start trial" button, and send 200 to 500 targeted visitors to it. Clicking the button can lead to an honest "we are opening soon" page. What matters is the click-through rate on the paid action, not signups for a newsletter.
Rough 2026 benchmarks from founders who publish their numbers: below 1 percent click-to-trial from cold targeted traffic usually means weak pain or wrong audience; above 5 percent is worth building for. Anything in between is decided by the quality of the interview data, which is why the two tests run together.
- Ask about the last time the problem happened, never about hypothetical futures.
- Record exact phrases customers use; they become your landing page copy.
- Charge money in the smoke test — free signups flatter every idea.
- Kill criteria in writing before the test starts, not after you like the results.
The 40 percent check
Once a product exists and people use it, Sean Ellis's survey question gives a hard number for product-market fit: ask active users how they would feel if they could no longer use the product. If fewer than 40 percent answer "very disappointed," the fit is not there yet. Rahul Vohra ran this loop publicly at Superhuman in 2017, segmenting responses and rebuilding the roadmap around the "very disappointed" group until the number crossed the line.
| Signal | Strength | Cost to get | Can be faked by politeness? |
|---|---|---|---|
| "Great idea!" | None | Free | Always |
| Email signup | Weak | One landing page | Often |
| 30-minute interview with stories | Medium | Your time | Rarely |
| Click on a paid plan | Strong | Page + traffic | No |
| Pre-order or deposit | Strongest | Checkout link | No |
When the answer is no
A failed validation is a cheap success: it returned two years of your life. The founders who suffer most are not the ones whose idea died in week two but the ones who kept a zombie alive on hope and small compliments. Run the tests, write down the numbers, and if strangers will not click a price, believe them. The next idea will benefit from everything this one taught you about where the pain actually sits.