Most new businesses do not fail because the idea was bad. They fail because nobody found out soon enough that the idea was expensive to sell. Two weeks of structured conversations and one small paid test costs very little and tells you more than six months of building.
What validation actually means
An idea is validated when a person who has the problem gives up something scarce to solve it: money, a deposit, a signed commitment, or a slot in their own calendar. That is the whole test. Everything else — survey answers, likes, “let me know when it launches” — is a signal about politeness rather than demand.
This matters because founders usually collect the wrong evidence. It is much more comfortable to ask whether an idea sounds good than to ask for a deposit. A yes to “sounds good” costs the other person nothing, so it tells you nothing.
Week one — get specific about the buyer
Vague buyers produce vague answers. Before you talk to anyone, narrow three things to the point where you could find a room full of these people:
- Who exactly. “Small businesses” is not a buyer. “Independent coffee shops with two to five sites that buy their own equipment” is a buyer you can actually find.
- The job they are trying to get done. Not the product you want to sell — the outcome. “Get the health inspection passed without closing for a week” is a job.
- What they do now instead. Every real problem already has a workaround, even a bad one: a spreadsheet, a friend, a manual process, doing nothing. Your competition is that workaround.
The five conversations, and what to ask
Ask about the past, not the future. People are accurate about what they did and unreliable about what they will do. A conversation that follows this order will produce usable evidence:
- “Walk me through the last time this happened.” Get the timeline, the people involved and the cost. Listen for emotion — frustration is a good sign that a problem is pressing.
- “What did you do about it?” Whatever they did next is your real competitor.
- “What did that cost you?” In money, hours, or lost customers. If the answer is “nothing much”, the problem is not urgent enough to sell against.
- “Who else has to agree?” For business buyers this is how you discover that your champion is not the decision-maker.
- “If this existed, what would you expect to pay for it?” Ask once, then stop talking. The number they say out loud is information even when it is wrong.
Do not pitch during the first three conversations. The moment you describe your solution, the other person starts being helpful, and helpful answers are nearly worthless for this purpose.
Week two — ask for something real
In the second week you stop researching and start testing. Pick whichever of these fits your situation; the test is the same in each case — does someone commit before the thing exists?
| If you plan to sell… | The cheapest honest test | What a pass looks like |
|---|---|---|
| A physical product | Pre-order at full price with clear delivery terms, or a refundable deposit | Any stranger — not a friend — pays before you order stock |
| A service to businesses | A paid pilot at cost, or a signed letter of intent naming a price and a start date | Two of five conversations end in a dated commitment |
| Software or a subscription | Founding-member pricing for a fixed term, paid annually, delivered manually at first | Payments arrive even though the product is a spreadsheet and a chat thread |
| Wholesale or importing | A small paid sample order, then one full order at normal terms | A buyer repeats the order without you chasing them |
| Anything with a hard-to-reach buyer | A waitlist with a refundable deposit and a published price | Deposits keep arriving after your first announcement, not just on day one |
Whatever the test, write the terms on one page and send it before you take money: what they are buying, when it arrives, what happens if it does not, and how to get a refund. Testing demand with vague promises creates a different problem — and it also produces the wrong answer, because vague promises are easy to say yes to.
Reading the result without fooling yourself
Two weeks ends with one of four outcomes. Only the first is a clear go.
- People paid, and some were strangers. Continue, but keep the scope small. The question has changed from “is there demand” to “how do we deliver this repeatedly”.
- People were interested but nobody paid. Treat this as a pricing, packaging or urgency problem rather than proof of demand. Test a smaller, cheaper, faster version — often the same idea cut down to one narrow use case.
- Nobody was interested. That is a good outcome and it cost you two weeks. Either the buyer is wrong or the problem is not pressing. Try a different buyer before abandoning the idea entirely.
- You never got to week two. Almost always this means you could not identify or reach enough buyers. That is a distribution problem, and it is worth solving before anything else — the best product in the world is worthless if you cannot find the people who need it.
Five ways founders fake validation
- Asking friends and counting their answers. Friends are biased toward encouragement by design.
- Counting social engagement as demand. A like costs one tap; a deposit costs money and trust.
- Asking “would you buy this?” Hypothetical questions get hypothetical answers.
- Moving the threshold after the results. If the target was two commitments, “people seemed keen” is not two.
- Building first because the conversation is uncomfortable. Building feels like progress and avoids the only test that matters.
The two-week timetable
| Day | Do this | Output you can point at |
|---|---|---|
| 1 | Write the buyer, the job, the workaround and your pass/fail threshold | Half a page, dated |
| 2–3 | Build a list of 20–30 people who fit the buyer profile | A list with names and a way to reach each one |
| 3–7 | Run five structured conversations using the questions above | Notes with direct quotes on cost, urgency and who decides |
| 8 | Rewrite the offer in the words your five buyers actually used | A one-page offer with a price |
| 9–12 | Ask every contact for the real commitment (deposit, pilot, intent) | Money in, or signed intent with a date |
| 13 | Deliver the first order manually | One unhappy surprise you did not expect |
| 14 | Compare the result to the threshold and write the decision down | Go, narrow, change buyer, or stop |
Frequently asked questions
How many customer conversations are enough to validate an idea?
For a first read, five structured conversations with people who fit your buyer profile is enough to find the pattern. Fewer than five and you cannot tell a real objection from one person's mood. More than ten before you have asked anyone for money is usually procrastination.
Is a survey as good as a conversation?
No, but it is useful in a different way. Surveys measure how common an opinion is; conversations reveal the reasoning and the workaround people already use. Run five conversations first, then use a short survey to check whether the pattern holds across a wider group.
What if nobody will pay a deposit yet?
Ask for the smallest real commitment instead: a signed letter of intent with a price and a date, a paid pilot at cost, or a slot on a waitlist with a refundable deposit. A polite “sounds interesting” is not a commitment, and it should not count toward your pass threshold.
How much should I spend during validation?
Assume nothing you spend is recoverable and keep it in the low hundreds, not the low thousands, of your own currency. The output of validation is information, so spend your time and your conversations rather than your capital.