How to Validate a Startup Idea Before Building: 7 Proven Steps

Last month, a founder pitched me a scheduling tool for dental clinics. He'd already spent 4,200 euros on a developer in Lisbon. When I asked how many clinic managers he'd spoken to, he said: "Two. But they both loved the idea."

That sentence is the most expensive mistake in early-stage startups, and it costs far more than 4,200 euros. How you validate a startup idea before building determines whether you spend the next eight months building something people applaud and never pay for. Polite enthusiasm is not validation. It's the sound of a door quietly closing.

What follows is the process I've landed on after watching too many founders (including myself) confuse a friendly conversation with market demand. No frameworks with invented statistics, no growth-hacking shortcuts. Just the sequence that actually tells you whether to build — and how to know when to walk away.

How to validate a startup idea before you write a single line of code

Validation is a narrowing exercise. You start with a belief ("clinics lose money on no-shows") and you try to kill it. Every week you fail to kill it, the belief gets slightly stronger. That's it. That's the whole method.

Most founders run this backward: they start by building a small version and calling it a test. But a prototype tests your execution, not the market. The market question comes first, and it's answered with conversations and commitment requests, both of which cost almost nothing.

Start with a falsifiable hypothesis

"There's an opportunity in healthcare scheduling" is not a hypothesis. It's a mood. Rewrite it as something that could be proven false:

• Who exactly: managers of dental clinics with 2 to 5 chairs in urban areas.
• What pain: they currently lose more than 8% of appointment slots to late cancellations each month.
• What they do today: they call patients manually, using a front-desk spreadsheet.
• What would prove me wrong: if 20 interviews show that no-show rates are under 3%, or that they've already bought a solution and are happy with it.

Notice the last bullet. Write it down before the first call. It's the only protection you have against your own optimism — and your optimism is the most dangerous variable in this entire process.

Pick a segment narrow enough to be wrong about

Here's where I'll take a firm position: broad validation is fake validation. If your idea "helps small businesses save time," you'll collect twenty different reactions from twenty different contexts and learn nothing. If your idea helps independent physiotherapists in a specific billing situation, ten conversations will either converge or scatter — and both outcomes are informative.

Narrow segments feel smaller than the opportunity you imagine. Good. You're not trying to size the market yet. You're trying to find out whether anyone has the problem badly enough to act.

How to talk to customers without collecting fake signals

The interview is where most validation efforts quietly fall apart. Not because founders don't talk to people, but because they ask questions that guarantee a pleasant answer.

I learned this the hard way. On an earlier project, I ran eleven interviews and came away convinced I had a winner. Every single person said they'd use it. Six months later, I had three users and a very expensive lesson about question design.

Ask about the past, not the future

Hypothetical questions produce hypothetical data. "Would you use a tool that does X?" is a question about a person's imagination, not their behavior. Instead, anchor every question in something that already happened:

1. "Walk me through the last time this problem came up. What day was it?"
2. "What did you do in that moment? Who else got involved?"
3. "What did that cost you — in time, money, or a client?"
4. "Have you looked for a solution before? What did you try, and what made you stop?"
5. "What's the workaround you're using right now?"

The pattern to listen for is specificity. People describing real pain give you dates, names, and small humiliating details. People describing imagined pain give you generalities and compliments. Compliments are the enemy.

How many interviews are enough?

You'll hear numbers thrown around — ten, twenty, fifty. The honest answer is that the number matters less than the pattern. In my experience, by interview seven or eight within the same segment, new conversations stop producing new information. That's your saturation point.

If you're at interview twelve and each person describes a completely different problem, stop. That's not a sign you need more data. It's a sign your segment is too wide.

How to test willingness to pay before the product exists

Interest is free. Payment is not. The gap between "I'd definitely use that" and "here's my card" is where most startup ideas actually live or die, and the only way to measure it is to ask for something real.

What counts as a real ask, ranked by how much signal it gives you:

• A pre-payment or deposit for early access. Refundable is fine. The refusal rate is the data.
• A signed letter of intent from a business customer, with a price and a rough timeline on it.
• A calendar commitment — a scheduled onboarding call, not a vague "let's talk in a few months."
• Time invested: filling in a detailed onboarding survey, importing their existing data.
• Reputation risk: introducing you to their boss or a peer in the same situation.

My rule of thumb, and I'll defend it: if nobody in your first fifteen qualified conversations will commit money or a calendar slot, you don't have a validated idea — you have a hypothesis with fans. That doesn't mean the idea is dead. It usually means the pain isn't urgent enough to justify a purchase, and urgency is what you're really selling.

What is a good conversion rate for a pre-sale test?

Be careful with benchmarks here, because they depend entirely on price and segment. What I can tell you from my own runs: on a €39/month B2B tool aimed at freelancers, roughly one in twelve cold conversations converted to a pre-payment. On a €400/month tool for agencies, one in six converted — a higher rate despite a much higher price, because the pain was operational and expensive.

Watch the ratio, not the absolute number. And watch the price sensitivity: if everyone says yes at your price, you're probably too cheap, and cheap prices attract customers with weak problems.

How to evaluate a startup idea without lying to yourself

Look, the hardest part of validation isn't gathering information. It's interpreting it honestly when you're emotionally invested in the outcome. You will find reasons to discount the negative signals. Everyone does.

The fix is mechanical: decide your thresholds in advance, in writing, before the first conversation.

Signal	Weak (keep digging)	Decent	Strong enough to build
Interviews showing the same concrete pain	Fewer than 5 of 15	6–9 of 15	10+ of 15, unprompted
People using a paid or manual workaround today	Under a third	Around half	Most of them, and they complain about it
Pre-payments or signed intents	Zero	2–3	5+ within a narrow segment
Unsolicited referrals	None	1	Several, without you asking

These aren't universal laws. They're the numbers I use, and they've kept me from two bad builds. Adjust them to your context — but adjust them before the interviews, not after.

How to validate a business idea when you have no network

This is the objection I hear most, and it's fair. You don't need a network; you need a channel. Cold outreach on a professional network, a short message in three relevant communities, a plain one-page site with a single call to action, or simply showing up where your segment already gathers.

What you cannot do is skip the conversation and call the idea validated because a landing page got clicks. A click costs the visitor nothing. It measures curiosity, and curiosity is nearly free.

The signals that should make you kill the idea

Nobody writes about this part, so I will. Killing an idea early is a success, not a failure. Here's what I treat as grounds for a hard stop:

• Your target customers describe the problem but have no budget line for it and no authority to create one.
• Everyone has a workaround they consider good enough — spreadsheets, a WhatsApp group, an intern.
• The pain is real but infrequent: twice a year isn't a business.
• Your best interviews are all outliers, and the mainstream of the segment shrugs.
• You keep reframing the segment to find people who say yes. That's not iteration, that's avoidance.

I killed one project after nine weeks of this. It stung for about a week. Then I realized I'd saved myself roughly six months and a chunk of savings, which is a trade I'd take every time.

What to actually build first

Once the signals clear your thresholds, you still don't build the full thing. You build the smallest artifact that delivers the core value to your first handful of customers — often a manual process behind a clean interface. Some founders I know ran their first three months entirely on a form, a spreadsheet, and their own labor. Nobody complained, because the value was in the outcome, not the software.

That phase tells you something no interview can: whether people keep using the thing after the novelty wears off. Retention is the only validation that survives contact with reality.

The uncomfortable truth about validating a startup idea is that it never feels conclusive. You will always want one more conversation, one more data point, one more confirmation. At some point you have to make a decision with 70% of the information you'd like — and the discipline of setting thresholds in advance is what lets you make it without flinching.

Which raises a question worth sitting with: if you're not willing to ask fifteen strangers for money before you build, what exactly are you expecting those eight months of development to change?