How to validate a business idea before you build

Write down your riskiest assumption

Every idea rests on a stack of assumptions, and they are not equally dangerous. Most of them break in ways you can recover from. One or two would end the business outright. Validation is the work of finding those one or two and testing them first, while a wrong answer still costs you a week instead of a year.

Write yours down under four headings:

• The problem. A specific person has this problem, and it bothers them often enough to act.
• The willingness to pay. They will hand over money, or something equally scarce like their time or their email, to make it go away.
• The channel. You can reach enough of those people at a cost the business can carry.
• The delivery. You can actually build and run the thing you are promising.

Now score each one twice: how badly would it hurt if this turned out to be false, and how sure am I really, on evidence rather than instinct. The assumption that is both high damage and low certainty is where you start. Everything else waits.

For most first-time founders the honest answer is not "can I build it". It is "will anyone want it enough to pay". Naming that out loud is the point of this step, because the natural pull is toward the assumptions you already feel good about. A week spent proving you can build the thing tells you nothing about whether you should.

Market validation and market research are different jobs

These two get used as if they mean the same thing, and mixing them up is why founders spend a month on a deck of statistics and still do not know whether to build.

Market research answers "is there a market here": how big it is, who else is serving it, what people pay today, how it is changing. It is desk work, it is largely reading, and it is answered with sources.

Market validation answers a narrower and more personal question: "will these specific people take a real action in response to my specific offer". No amount of research answers that, because it has never been tested. Only a real offer put in front of real people answers it.

You need both, and in that order. Research tells you whether the market is worth entering and stops you validating an idea inside a market that cannot support it. Validation tells you whether your version of the offer earns a response. If you have not done the research half yet, start with market research for startups and come back here.

If you are comparing tools rather than doing the work. Plenty of people searching this phrase want a shortlist, not a method. That is a fair thing to want, and it is a different page: see the AI idea validation tools compared instead. The rest of this guide assumes you want to run the test yourself.

Talk to ten real potential customers

Not friends, not family, not anyone with a reason to be kind to you. Find ten people who match the customer in your riskiest assumption, and talk to them about their problem rather than your solution.

Ask about the past, not the future. What people say they would do is close to worthless; what they did last month is evidence.

• What do you do about this today?
• Walk me through the last time it happened.
• What have you already tried, and why did you stop?
• What did it cost you when it went wrong, in money, time, or someone being annoyed with you?
• Who else is involved when you fix it?

Do not pitch. The moment you describe your product, the conversation stops being research and becomes a politeness test that you will pass and learn nothing from.

What you are listening for is whether the problem is a painkiller or a vitamin. A painkiller shows up as a workaround they already built, money they already spend, or visible irritation when they describe it. A vitamin shows up as "yeah, that would be nice". Nice does not pay.

Two results are worth as much as ten good conversations. If you cannot find ten people who have the problem, you have learned the channel assumption is wrong before spending anything on it. And if all ten describe a different problem to the one you assumed, you have just been handed a better idea for the price of a few phone calls.

Run a real demand test, not a survey

People are polite in surveys and honest with their email address. A survey measures opinion; a demand test measures behavior. Only one of them predicts a business.

Put up a page that describes the product and the result it produces, as if it already exists, with exactly one thing to do: join the waitlist, pre-order, or book a call. Then send a small, honest amount of traffic from where your customer actually is, and measure the rate at which the right people take that action.

Three things decide whether the number means anything:

1. The traffic has to be the real audience. A hundred visitors from a founder community will sign up for almost anything. A hundred from the place your customer already spends time will not.
2. The action has to cost something. An email is the minimum. A deposit, a booked call, or a card on file is far stronger evidence, because the price of saying yes is what makes yes meaningful.
3. The offer has to be honest. Describe what you intend to build. Never take money for something you have decided not to make, and say plainly that it is early.

The full mechanics, including what to put on the page and how much traffic is enough, are in how to run a fake-door test. A page that the right people ignore is not a failure. It is a clear, cheap answer that would have cost you a year to get any other way.

Set your pass mark before you look at the result

Decide the number that would make you continue before you run the test, and write it down where you cannot quietly edit it later.

Something like: if at least 8 of every 100 visitors from my target audience leave an email, I build the next piece.

The number matters less than the timing. Set it in advance and the test can tell you something you did not want to hear. Set it afterwards and you will find a way to read almost any result as encouraging, because by then you are attached to the idea. This is the single cheapest protection against fooling yourself in the whole process, and it costs one sentence.

Be careful that the rate is measured against the right denominator. Eight signups from a hundred targeted visitors is a signal. Eight from a thousand mixed visitors, most of whom were never your customer, is noise wearing the same number.

What each result actually means

You cleared the bar. Build the next smallest testable piece, not the whole product. You have evidence that this audience responds to this promise, which is not yet evidence that they will keep using it or pay again. The next question is retention, and the fastest way to ask it is a product market fit survey once people have actually used the thing.

You missed it badly. Almost always the problem is the audience or the promise, not the idea. Before abandoning it, change one variable and rerun: same offer to a sharper audience, or a sharper promise to the same audience. Two clean misses in a row against your real customer is a genuine answer, and it arrived for the price of a week.

You landed near the line. This is the most common outcome and the most dangerous, because the temptation is to run it again until it passes. Do not. Go back to the conversations instead: a near miss usually means the problem is real but your promise describes it in words the customer does not use.

None of the three outcomes is a failure, and only one of them is a green light. The point of validating a business idea is not to be told yes. It is to find out which it is while it is still cheap to be wrong.