◢Most founders skip the part that actually matters
Learning how to validate a startup idea is the most boring, most skipped, and most important thing a founder will do. It is also the cheapest insurance you will ever buy. We are Sameer and Ankit, and we have watched this movie a hundred times.
The plot is always the same. A founder gets a great idea. They love the idea. So they build it for six months, launch to silence, and then learn the lesson the expensive way.
Here is the stat that should scare you straight. CB Insights studied 431 startups that shut down and found poor product-market fit was a top reason at 43 percent of failures. Cash running out tops the list, but that is the symptom. Nobody wanting the thing is the disease.
We ran a growth agency and built our own "obvious" products before doing the homework. Nobody pays us to say any of this. So here is the honest, cheap, slightly ruthless way to find out if your idea is real before you waste a single weekend building it.
◢What does it actually mean to validate a startup idea?
To validate a startup idea means proving two things with evidence: that a real problem exists, and that real people will pay to solve it. It is not about polishing your pitch or counting likes. It is about collecting hard signals (cash, signups, clicks) from strangers who owe you nothing.
Validation lives before the build, not after. The frame we use comes straight from lean startup thinking: chase problem-solution fit first, then worry about product-market fit later. Antler puts it plainly, that you must prove a deep customer problem in an underserved market before you build a polished product.
Most founders confuse the two. Problem-solution fit means people want your proposed fix. Product-market fit means you shipped it and they stick around. Steve Blank built a whole customer development model around this order, and it is the spine of everything below.
So what to cut here? The instinct to build first. Your idea is a guess until a stranger proves otherwise. Validation is just structured guess-killing, and the faster you kill the bad guesses, the more runway you keep.
◢How do you validate the problem before the product?
You validate the problem by talking to real people about their lives, not your idea. Run 15 to 30 honest conversations with people in your target market. Ask what they do now, what is hard, and when it last hurt. The second you pitch, the data goes bad.
This is the entire lesson of The Mom Test by Rob Fitzpatrick. The book's three rules are simple: talk about their life not your idea, ask about specific past behavior not future hypotheticals, and listen more than you talk. People lie to be nice. Good questions stop them.
Y Combinator says the same thing in fewer words. Their core advice is to make something people want, and you only learn what people want by talking to them constantly. Ask "what is the hardest part?" and "what did you do last time?" Never ask "would you buy this?"
A few field notes from us:
- Interview like a doctor. Ask where it hurts before you reach for the medicine. If you describe your cure first, they will agree just to be polite.
- Chase the pain, not the praise. "That sounds cool" is a death rattle. "I literally pay for a hacky workaround right now" is gold.
- Watch for money already moving. The strongest signal is a budget that already exists. People who pay to limp through a problem are people who will pay to actually fix it.
When do you stop interviewing? When you hit saturation. The customer discovery playbook says you are done when three people in a row tell you nothing new and you can predict the next answer.
◢How do you test real demand without building anything?
You test demand with a cheap experiment that forces a real action: a smoke-test landing page, a fake door, or a pre-sale. Words are free, so words lie. Make people spend something they hate parting with (an email, a click, a card number) and the truth shows up fast.
This is the step that separates founders who validate from founders who just feel validated. Below are the three demand tests we actually use, cheapest first.
The smoke-test landing page
Build one page that describes the product like it exists, then add an email signup or a pre-order button. Drive a few hundred visitors and watch the conversion rate. A landing page smoke test measures unmet demand in your niche before you write a line of code.
Two non-negotiables. Pick your success number before you launch, because deciding afterward is just rationalizing. And aim for 200 to 500 visitors so the result is signal, not noise. Below that, you are reading tea leaves.
Buffer is the textbook case. Joel Gascoigne built a tiny two-page site, added a pricing page before the email form to test willingness to pay, and went from idea to paying customers in seven weeks. The pricing page was the clever part: it filtered tire-kickers from buyers.
The fake door test
A fake door is the same trick inside an existing product. You add a button, a feature card, or a pricing tier for something you have not built, then count the clicks. Amplitude describes fake door testing as tracking clicks on a non-functional element to measure interest before you commit engineers.
The fake door smoke test guide suggests running it for 7 to 14 days and gathering at least 300 sessions per audience. Be decent about it: when someone clicks, show an honest "coming soon" and offer a waitlist or a chat. Do not burn trust for a data point.
The pre-sale or waitlist
The strongest signal is money or a committed signup before launch. Dropbox is famous here. Drew Houston posted a three-minute demo video of a product that barely existed, and the waitlist jumped to 75,000 signups overnight. No app, just proof of demand.
Robinhood did it with a referral waitlist and a landing page, hitting one million signups before launch and using that demand to help raise funding. You do not need a million. You need enough committed strangers to prove the problem is worth your year.
For wiring up that landing page and capturing leads cleanly, our inbound lead capture recipe shows the lean stack, and our roundup of the best free tools for startups covers the no-cost picks. You can run every test in this section for roughly the price of a domain.
◢What numbers actually count as a "yes"?
A "yes" is a real action at a rate you set in advance, not a vibe. For a smoke-test page, that means hitting your pre-chosen conversion target with a few hundred real visitors. For interviews, it means a clear, repeated pain that people already spend money or time fighting. Set the bar before you look, or you will move it.
The classic mistake is grading your own homework after the fact. Forum Ventures, in its idea validation playbook, pushes founders to define success criteria up front so you cannot wriggle out of a bad result. We do the same: write the number on a sticky note before the experiment runs.
Later, when you have a product in users' hands, the gold-standard demand metric is the Sean Ellis test. You ask how people would feel if they could no longer use your product, and the 40 percent benchmark says you likely have product-market fit when 40 percent answer "very disappointed." That is a post-launch tool, but it is the same idea: measure real attachment, not applause.
What to cut: the temptation to count followers, likes, and "great idea!" replies as validation. None of those cost the giver anything, so none of them prove demand. A single pre-order beats a thousand thumbs-up.
◢How long should validation take, and when do you stop?
Validation should take weeks, not months. A round of 15 to 30 interviews plus one demand test usually fits inside three to five weeks. You stop when the evidence clearly points one way: real pain plus real action means go, polite nods plus zero signups means pivot or pass.
Speed is the whole point. Every week spent validating is a week not spent building the wrong thing. The cost of a smoke test is a domain and some attention. The cost of skipping it is the six months and the silent launch we keep watching founders walk into.
Three honest endings are all on the table. You validate and build with conviction. You invalidate and save yourself a year. Or you learn the real problem was next door, and you pivot toward it. All three are wins, because all three are cheap.
Once you do build, keep validating with real usage. Our product analytics stack recipe shows the lean way to watch whether people actually stick, so you catch a weak idea before it eats your runway. Validation does not stop at launch; it just changes tools.
◢Conclusion: kill bad ideas while they are still cheap
Validation is not a phase you survive before the fun starts. It is the cheapest, fastest tool you have for not wasting a year of your life. Prove the problem with honest interviews, prove the demand with a smoke test or a pre-sale, and set your pass-or-fail number before you peek.
The takeaways are simple. People lie to be nice, so measure actions not opinions. Build only after strangers vote with their email or their wallet. And remember that 43 percent of dead startups died because nobody wanted the thing, a fate a few weeks of testing usually prevents.
Once your idea earns a real "yes," the next problem is getting to market without drowning in tools. That is the entire point of our go-to-market recipes: wire the lean stack, skip the bloat. You can even tally what a full stack would cost you with our free stack cost calculator before you commit a dime.
Want the no-sponsor playbooks we actually run, plus the tools we tell founders to cut every week? Subscribe to the newsletter. We bash the bloat so your runway lasts.
◢FAQ
How do you validate a startup idea before building it?
Validate in two steps. First, confirm the problem is real by interviewing 15 to 30 people in your target market about their actual past behavior, not your idea. Then test demand with a cheap experiment: a smoke-test landing page, a fake door button, or a pre-sale. If strangers give you their email, their click, or their money, you have a signal. If they only say nice things, you have nothing. Build after the demand test, not before.
How many customer interviews do you need to validate an idea?
Most founders need 15 to 30 focused interviews in one customer segment. Rob Fitzpatrick, who wrote The Mom Test, suggests 20 to 30 discovery chats. Steve Blank pushes for 50 to 100 across discovery and validation. The real finish line is saturation: when three people in a row tell you nothing new and you can predict the next answer, stop. The goal is patterns, not a magic number.
What is the cheapest way to test demand for a startup idea?
A smoke-test landing page is the cheapest demand test. You build one page describing the product, add an email signup or a pre-order button, and drive a few hundred visitors to it. The conversion rate tells you if real demand exists before you write any code. Fake door tests and short pre-sales work the same way. All three cost more in attention than money, and all three beat guessing.
What is the difference between problem-solution fit and product-market fit?
Problem-solution fit means you have proven a real, painful problem exists and that people want your proposed fix. Product-market fit means you shipped a product, customers stick around, and growth is healthy in a big-enough market. Problem-solution fit comes first and is what idea validation is really about. Sean Ellis suggests you have product-market fit when 40 percent of users would be very disappointed to lose your product.
What is a smoke test or fake door test?
A smoke test is a landing page for a product that does not fully exist yet, built to see if people will sign up, pre-order, or book a call. A fake door test is the same idea inside an app: a button or pricing tier for a feature you have not built, measuring how many people click. Both prove intent with real actions instead of opinions. Run them for 7 to 14 days and aim for at least a few hundred sessions.