To validate a business idea, have twenty real conversations with people in your target market in the next two weeks — before you write a single line of code. Not surveys. Not polls. Not friends. Twenty is not a magic number — it's the number where patterns become legible. At ten conversations, you can convince yourself of anything. At twenty, the pattern is undeniable.
Here is the exact sprint: who to talk to, the five-section script, the questions that poison your data, and how to read the results honestly.
The six-month mistake validation exists to prevent
Picture a founder we'll call Marc — an archetype, a composite of the most common founder story I know. Head of product at a logistics company in Rotterdam, with a genuinely good idea: a scheduling tool for freight forwarders. He started building with AI as his copilot — and didn't talk to a single freight forwarder before writing his first line of code. He told himself he'd start the conversations once the product was ready.
Six months in, he had a beautiful working product. He sent the link to twenty freight forwarders. Three replied. One tried it for ten minutes and wrote back: "Looks great, but we already use TMS-Connect for this and it's bundled with our customs software, so I don't think we'd switch." The switching cost wasn't his price — it was the price of unbundling from a workflow with eighteen other things wired into it. He killed the project three weeks later: four thousand euros on tools, two hundred evenings, sixty weekends. Two hours of conversation in week one would have prevented all of it.
What validating a business idea is NOT
Validation is not a survey. A survey gets you a hundred lukewarm responses from people with no skin in the game, clicking radio buttons at red lights. Surveys tell you what people think they think — not what they actually do, and what people actually do is the only thing that matters.
Validation is not a Twitter poll — that's entertainment. And it's not asking your friends if your idea is good. Your friends love you, they don't know your market, and they'll tell you it's brilliant because that's what friends do. Don't put them in that position.
Validation is twenty to fifty real conversations with people who are in your actual target market, have the actual problem, and have actual money to spend on solving it. It's not complicated — just uncomfortable, and the discomfort is exactly why doing it gives you an unfair advantage over everyone who skipped it.
The 20-Conversation Sprint: step by step
Define who you're talking to. Not "anyone who might want this." Something narrow: founders of B2B SaaS companies between ten and fifty employees. You can always broaden later. You cannot validate "everyone" — there are no patterns at that level of generality.
Find them. LinkedIn, industry communities, the conferences your competitors attend. Send a message of three sentences maximum: who you are, that you're researching the problem space (not selling anything), and a thirty-minute ask. About one in three will respond — so send sixty messages to get twenty conversations.
Run the same script every time. You're looking for the pattern across answers, and that requires consistency. Five sections:
| Section | The question | Time | What you're listening for |
|---|---|---|---|
| 1. Warm-up | "Tell me about your role and your team." | 2 min | They speak first, they get comfortable |
| 2. Current state | "How do you currently handle [the problem area]?" | 5 min | Tools and what they cost; workflows and how often they break |
| 3. Cost of the problem | "What's the impact of this not being solved?" | 5 min | Hours or euros = real problem; a shrug = theoretical, and theoretical problems don't get paid for |
| 4. Alternatives | "What have you tried? What did you like and not like?" | 5–10 min | People who've actively tried to solve it are buyers; people who haven't are not — the single highest-signal question in the script |
| 5. The close | "What would you expect to pay for it?" Then: "Would you be one of the first ten customers?" | — | Their price point, unprompted; then yes / maybe / no |
Log everything. One spreadsheet, one row per conversation: name, role, current solution, cost of the problem, price expectation, yes/maybe/no. Write down direct quotes. By fifteen conversations, the spreadsheet is telling you a story — and unlike your gut, it doesn't care about your feelings.
The questions you should never ask
Three questions poison validation data: "Would you use a tool that does X?" (leading), "Do you think this is a problem?" (yes-or-no, leading), and "How important is this to you?" (forces a politeness answer). None of the five script questions ask people to predict their own future behavior. They ask people to describe their actual past behavior — the only honest predictor of future behavior.
The follow-ups matter as much as the mains. When they describe a frustration, ask "tell me about the last time this happened" — specific incidents are gold; abstractions are noise. And after each question, clamp your mouth shut. Nervous founders fill silence by pitching, and once you've told someone what to think, their answer is worthless.
Reading the signals: red, yellow, green
Red — kill it. Most people don't have the problem. Or they have it but can't quantify the cost. Or they say it's "interesting" but can't name what they'd pay. Or everyone would "probably use it" but nobody would pay for it.
"Nice idea" is the most dangerous phrase in the founder vocabulary because it sounds like validation but means the exact opposite. It means: I'd accept this if it were free, but I'd never go out of my way to find it. Free products are not businesses.
Yellow — proceed cautiously. Most people have the problem, some quantify the cost, price expectations are scattered, and you got two or three soft commitments. This is the most common outcome. Do another ten conversations with a tighter customer definition; the pattern will sharpen.
Green — build it. Most people have the problem, can quantify it in real money or hours, have actively tried tools they don't love, price expectations cluster in a range you can charge profitably, and at least five people explicitly want to be early customers. Bonus green light: they're already paying for something that almost solves it. That's the easiest sale a solopreneur can make.
The presell test: money is the only honest signal
The conversations tell you the problem is real. The presell tells you your solution is. A presell is money changing hands before you build: a one-page landing site (AI can build it in an hour), a fifty-percent early-bird discount on what you'll charge later, and a Stripe checkout, with a clear note that delivery is in eight weeks. Email the five to ten people who said they'd be early customers. If three out of ten pay, you have a business. If zero pay, they were polite, not customers — and your spreadsheet was lying. Money is the only signal that survives politeness.
It's about three hours of work, and it's the single highest-leverage activity in the whole methodology. Only then do you build — and the build itself now costs around €340 a month in tools, so validation, not capital, is the real gate. If you want the guided version of this sprint, that's exactly what the Accelerator walks you through; the full framework is in the book.
Frequently asked questions
How many customer interviews do you need to validate a business idea?
Twenty to fifty. At ten you can convince yourself of anything; at twenty the pattern is undeniable; at fifty you're refining what you already know. Five conversations is noise — don't start building off three optimistic calls.
What if nobody responds to my outreach?
Cold outreach has a low response rate for everyone — about one in three replies is normal, which is why you send sixty messages to get twenty conversations. Below one in twenty after twenty messages? Fix the message (shorter, more specific), the targeting, or the channel. And if you can't get any conversations after thirty messages across channels, that silence is itself a signal: your market doesn't care enough about the problem to take a half-hour call about it.
Should I pitch my idea during validation calls?
No. Ask about their current situation, listen more than you talk, and let them fill the silence. The only forward-looking moment is the close: what they'd expect to pay, and whether they'd be one of the first ten customers.
What should I do if validation fails?
Kill the idea and be grateful. Most founders who validate properly discover their first idea isn't the one — that's not a failure of validation, it's the entire point. Two weeks of conversations is the cheapest tuition you'll ever pay. Start over on Monday with everything you learned about the market from the dead idea.