How to find out if anyone will pay, before you build

Most side projects fail on demand, not execution — and demand is testable in weeks for almost nothing. The only evidence that counts is someone handing over money before the thing exists.

The most expensive mistake in building anything is spending a year on something nobody wanted. It is expensive not because of the money but because of the time, and time is the resource you have least of if you have a job.

The good news is that demand is testable, cheaply, in weeks. The bad news is that almost every method of testing it gives a false positive, and the one that works is uncomfortable.

Why the usual tests fail

Asking people. “Would you use this?” reliably returns yes. People are kind, the question costs them nothing, and imagining a purchase is not the same faculty as making one. Enthusiastic verbal feedback is the single most misleading signal available.

Counting interest. Sign-ups, followers, waitlist entries, likes. All free to give. A thousand people who cost nothing to acquire tell you approximately nothing about whether ten would pay.

Market size research. A large market does not mean a reachable one, and it says nothing about whether you can reach it at a cost below what they pay you. The advertising auction decides that, not the size of the category.

Competitor existence. Useful — it proves demand exists — and it says nothing about whether there is room for you or whether the incumbents are profitable.

The common defect: none of them involve anyone giving up anything.

The only test that counts

Someone pays before the thing exists.

Not pledges. Not “definitely, when it’s ready.” Money, transferred, for something not yet built.

That is uncomfortable to ask for, which is exactly why it is informative. The discomfort is the mechanism — you are asking someone to make a real decision, and real decisions are the only data.

Practically, this looks like:

Sell the service manually first. Before building software that does something, do it by hand for a paying customer. If nobody will pay for it done manually, they will not pay for it automated. This is slow, unscalable, and the fastest way to learn whether the thing has a market.

Take deposits. Pre-orders, a discounted early price, a paid pilot. A small amount is sufficient — the size does not matter, the transfer does.

Sell the outcome, not the artefact. People pay for problems solved. Ask what they would pay to have the problem gone, not what they would pay for your thing.

What a real signal looks like

They pay without a discount being the reason. If the only interest is at 90% off, you have found bargain hunters.

They ask when they can have it. Impatience is the strongest indicator there is.

They tell someone else. Unprompted referral from a first customer is worth more than a hundred sign-ups.

They come back. For anything repeatable, the second purchase is the real validation. The first can be curiosity or politeness.

What a false signal looks like

Friends and family buying. They are buying you, not the thing. Exclude them entirely from your reading of the data.

One large enthusiastic customer. Could be a market; could be one unusual person. You need several unconnected strangers before you have evidence.

Praise from people in your industry. Peers evaluate whether it is clever. Customers evaluate whether it solves something. These correlate poorly.

Your own conviction. The strongest and least reliable signal in the set. You are the one person guaranteed to believe it.

The awkward part

Most people avoid this test, and the reason is worth naming: while you have not asked, the idea can still be good. Once you ask and nobody pays, you know.

So the building phase is comfortable and the asking phase is not, and people extend the comfortable phase indefinitely — polishing, adding features, redesigning — because it defers the verdict. A year later they have something excellent that nobody wanted, and they blame execution.

Asking early is not confidence. It is the willingness to be told no while it is still cheap to hear it.

What to actually do this month

Write down what you would sell and to whom. One sentence each. If you cannot name the buyer specifically — not “small businesses” but the actual kind of person — you are not ready to test yet, and that is itself the finding.

Find ten of them and ask for money. Not for feedback. For money, for something specific, at a real price.

Set the bar in advance. “If three of ten pay, I build it. If none do, I change the offer or the audience.” Deciding after you see the result is how people talk themselves into building it anyway.

And if nobody pays, change one thing. The offer, the price or the audience — not all three. Usually the audience is wrong before the idea is.

Why this is the highest-value habit

Everything else on this site is about which channel to build and what it will look like. This is the thing that determines whether the attempt was worth making at all.

Two evenings of asking strangers for money saves a year of building. And the people who succeed at this are not more talented — they found out sooner, which meant they got more attempts. Given how these fields are distributed, attempts are the resource that matters.