How to validate demand before building
Guide · Founder · 4 min read · last verified 2026-07-27
Demand validation is the work of collecting evidence that people want something before you spend months building it. The key word is evidence, and evidence is graded: it gets stronger as it costs the person providing it more to produce. A kind comment costs nothing. An hour of someone's time costs something. A commitment with consequences costs a lot. Validation is the practice of climbing that ladder deliberately — treating each rung like a deeper core sample, drilled further below the surface enthusiasm and closer to the bedrock of what buyers will actually do.
What counts as evidence, and what does not
Start by disqualifying the popular non-evidence. Encouragement from friends is not evidence; friends are usually optimising for the relationship, not for truth. Survey answers about future behaviour are weak evidence; people tend to answer surveys aspirationally, describing the person they would like to be. A long waitlist gathered from a vague promise is weaker than it looks, because joining it cost nothing. Even a competitor's existence, often cited as proof of demand, only proves that someone else believed demand existed — a belief, not a measurement.
What survives disqualification is a short list: questions buyers are already asking without your prompting, accounts of what buyers have already done about the problem, and commitments that carry a real cost. Those three form the rungs below.
Rung one: the questions buyers already ask
The cheapest genuine evidence is the question trail buyers leave behind. People tend to ask questions when something blocks them, so a standing volume of questions about a problem is standing demand for answers to it — and, often, for products that make the answers unnecessary. Look for those questions in community threads, sales-call notes, support archives of adjacent products, and the queries people type into search engines and AI assistants. Which surfaces expose those questions keeps shifting — the answer engines currently visible are examples, observed and changing — but the questions themselves are the durable signal, which is why evidence-first platforms like Magrios organise research around them rather than around any single channel.
Reading the trail takes judgment. You are looking for questions that recur, that come from people matching your intended buyer, and that reveal an attempt already underway — "how do I make X work with Y" signals more than "is X good". Choosing which of these questions to anchor your research on is its own craft, covered in How to choose your first buyer questions.
Rung two: what people say about the past
The middle rung is the interview — but only a particular kind. Interviews about the future tend to mislead, because hypothetical questions invite hypothetical selves to answer. Interviews about the past tend to be harder to distort: what did you try, what did it cost you, what broke, what did you do next. A buyer who has already assembled a spreadsheet-and-intern workaround for a problem is handing you validation; a buyer who says your idea sounds great is handing you politeness. The technique for keeping these conversations clean — question phrasing, silence, separating notes from interpretation — is its own discipline, laid out in How to interview customers without leading them.
Rung three: what people do when it costs them
The deepest sample is behaviour under cost. A refundable deposit tends to tell you more than any number of survey responses, because it asks the buyer to consult their actual priorities rather than their imagined ones. Cost does not have to be money: a signed pilot agreement, a scheduled onboarding with their team's time on the calendar, an introduction to their boss, access to their data — each is a commitment a merely-polite person tends to avoid. The pattern to watch is the conversion between rungs. Many enthusiastic interviewees and few committers is a common shape, and it usually means the problem is real but not urgent, or urgent but not yours to solve.
Reading the say-do gap honestly
The gap between what people say and what they do is not noise to be averaged away; it is the finding. When talk is warm and behaviour is cold, believe the behaviour, then use the talk to understand why the behaviour is cold. Sometimes the blocker is price expectations, sometimes a missing integration, sometimes the fact that the person who feels the pain is not the person who signs. Founders who average the two signals tend to build for the enthusiastic composite in the middle — a buyer who does not exist. Founders who interrogate the gap tend to find either a fixable objection or a timely reason to stop.
When to stop validating and build
Validation has diminishing returns, and it can become its own comfortable procrastination. A reasonable stopping rule: you can name the buyer precisely, you have heard the same past-tense problem story repeatedly from people who match that description, at least a few of them have accepted a real cost to move forward, and you know which assumption would have to be false for this to fail. At that point, write the assumption down and build the smallest thing that tests it. Your early market motion then becomes an extension of the same evidence habit — the first days of it are described in What to do in your first week in a new market — and the sizing questions investors will ask are easier to answer honestly if you have read What founders get wrong about market size. The ladder does not retire after launch; it just gets climbed with paying customers on it.