“We’ve validated the idea.” I hear that sentence from founders more than almost any other, and it usually comes with evidence. Thirty people loved it. Two hundred joined the waitlist. In a survey, 80% said they would use it, and a mentor called it a great idea.
It sounds like progress. Most of the time, it isn’t.
When CB Insights reviewed 101 startup post-mortems, 42% of founders said they built something nobody wanted. It was the most common reason for failure in the study. It is also the one cause founders can almost entirely avoid before writing a line of code.
Founders collect opinions about a product that doesn’t exist yet and call it validation. Then they skip the work that would tell them the truth.
Why the yes misleads you
Look at that evidence again. None of it describes something that already happened. It’s all opinion about a future that hasn’t arrived.
People are kind. Ask someone what they think of your idea, and you haven’t asked a research question. You’ve asked them to choose between honesty and your feelings. Most pick your feelings.
Founders also ask leading questions without noticing. “Wouldn’t it be useful if this happened automatically?” Nobody has ever said no to that.
Then there’s cost. A waitlist signup takes an email address and four seconds. A yes that costs nothing proves nothing.
And most founders go looking for a yes. You may have left a job for this idea. So you ask in a way that makes yes easy to say, and once you hear it, you stop asking. That’s confirmation bias, and being clever doesn’t protect you from it. The only defence is a method that stops you from steering the answer.
You can’t validate an idea, only a problem
Your idea is a guess at a solution. Until you build it, nothing in the world can check it. Asking people about it means asking them to imagine a product and predict how they’d behave around it. We’re all bad at that. Ask anyone in January how often they’ll go to the gym in February.
The problem is different. It’s already happening, whether your startup exists or not. Someone is losing an hour a week to it, or paying for a clumsy fix. That’s a fact, and facts can be checked.
That is what customer discovery is for. You hold structured conversations with people who have the problem, and find out whether it’s real, how much it hurts, how often it happens, and whether they already spend time or money on it.
You never need to mention your idea, and you shouldn’t. The moment you describe your solution, people stop telling you about their lives and start reacting to your pitch. Afterwards, you can’t tell the two apart in your notes.
Ask about the last time, not the next time
People don’t lie about their future behaviour, but they’re bad at predicting it. “Would you use this?” gets you a sincere description of the person they intend to be, and you can’t build a business on that.
The past left evidence: spreadsheets, invoices, calendar entries, irritated messages to colleagues. Three questions do most of the work.
- Tell me about the last time this happened.
- What did you do about it?
- What did that cost you, in time, money or something you gave up?
If they can’t remember the last time, the problem isn’t painful enough to be memorable. If they remember it but did nothing, they’re living with it. But if they remember it in detail and built a spreadsheet at 11pm to cope, you’ve found something real. People don’t build workarounds for problems they don’t have.
Define your customer by behaviour, not demographics. “Small business owner” isn’t a customer. “People who send at least five invoices a month and chase the late ones themselves” is.
Leave out anyone who wants you to win, including friends and family. They’ll tell you what you want to hear.
Read the pattern, then decide
For one clearly defined customer type, the same themes usually repeat within eight to 12 conversations. Stop when new conversations stop surprising you.
The strongest signal is the same workaround, invented separately by people who have never met. When two strangers have built the same ugly spreadsheet, you’re looking at a market. Count how many people said something, not how loudly one person said it.
Then make an honest call. If the problem is real, painful, frequent, and already costing people, persevere, and build the smallest thing that tests whether your solution helps. If it’s real but not the problem you expected, or it belongs to a different customer, pivot and take what you learned with you.
If nobody is bleeding, go and find a problem that hurts. Learning that in a few weeks, rather than after 18 months and your savings, is the cheapest good news you’ll ever get.
Decide what a no looks like first
The hard part is making the call before you go looking for one more yes to justify what you already wanted to build.
So before your first conversation, write down what evidence would tell you to stop. The version of you who has done twelve interviews will be more attached to the idea than the version of you reading this. That person shouldn’t be the one setting the bar.
This week, find five people who have the problem and book three conversations. Ask about the last time it happened, and don’t mention your idea.