You know people will buy your product when someone gives up something they care about to get it. Money, usually. Sometimes time, or access, or a signature that commits their budget. Everything short of that is a costume worn by a maybe. The whole skill of testing demand before you build is learning to rank the evidence honestly, then refusing to act on the weak stuff.
Opinions are the cheapest thing a person can give you
Ask a friend whether your idea is good and you have asked them to protect the friendship, not to assess a market. They will say yes. They will say it with detail and enthusiasm, which is what makes it dangerous, because detail feels like evidence.
The same trap shows up with confident-sounding tools. TechCrunch reported that hikers had to be rescued after planning a trip with Google Gemini, with the sheriff's office saying they had been advised to bring far less food and water than the group required (TechCrunch). The advice was fluent. It was also wrong, and nobody bearing the cost of being wrong was the one giving it.
That is the test to apply to every piece of feedback you collect. Ask who pays if this opinion turns out to be wrong. If the answer is only you, it is not evidence.
The evidence ladder, ranked
Here is the ladder, weakest rung first. Each step up costs the buyer more, and therefore tells you more.
- Opinions. "That's a great idea." Costs the buyer nothing.
- Email sign-ups. Costs them thirty seconds and a throwaway address.
- Waitlists with a qualifying question. Costs them a few minutes and some honesty about their situation.
- Pre-orders and letters of intent. Costs them a decision, and sometimes an internal conversation with their boss.
- Deposits. Costs them money they cannot easily get back.
- Paid pilots. Costs them money, calendar time and reputation inside their own organisation.
Most people testing an idea live on rungs one and two, collect a pile of encouraging noise, and build for six months. Then they discover that a mailing list is not a customer list.
Rungs one to three: opinions, sign-ups and waitlists
You cannot skip conversations entirely. You just have to run them so the answers are worth something. Stop asking "would you buy this" and start asking about the last time the problem actually bit them. What did it cost? What did they do instead? What did they already pay someone to fix it? Past behaviour is the only part of an interview that behaves like data.
Email sign-ups are a small step up because they involve a tiny cost. The problem is that people sign up for things out of curiosity, politeness and boredom. A list of two hundred addresses from a landing page tells you your headline is interesting. It does not tell you the price is acceptable, the problem is urgent, or the buyer has budget.
Waitlists get useful the moment you add friction on purpose. Ask for the company, the role, the size of the problem, and what they are doing about it right now. Half your sign-ups will drop off. Good. The half who answer are the only half you were going to learn anything from.
Treat rungs one to three as targeting, not proof. They tell you who to point the real test at.
Rungs four to six: pre-orders, deposits and paid pilots
A pre-order is the first rung where the buyer takes on risk. They are paying for something that does not exist yet, on your word. Even a small pre-order price separates the people who like the idea from the people who have the problem.
A letter of intent is the business-to-business version, and it is more useful than it looks. It rarely carries legal weight, so do not treat it as a contract without asking someone qualified, but the act of getting a manager to put their name on a document forces them to have the internal conversation that reveals whether budget exists.
A deposit is stronger again, because refund friction is real friction. When someone hands over money that is awkward to claw back, they have made a decision they will now defend to themselves.
The paid pilot is the top of the ladder and the only rung that answers every question at once. A buyer pays you to solve the problem, once, manually, for real. You find out whether they will pay, what they will pay, whether the solution works, and whether delivering it is a business you actually want. Nothing you build later will teach you as much.
What each rung costs you to run in a week
You have a full-time job. So here is the ladder again, priced in your evenings rather than in dollars.
- Interviews: one week. Ten to fifteen conversations if you already know the audience. Longer if you are cold-outreaching strangers.
- Landing page with sign-ups: two or three evenings. Copy, a form, and a way to send traffic. The traffic is the hard part, not the page.
- Qualified waitlist: same build, plus one evening to write the qualifying questions properly. Cheapest upgrade on the whole ladder.
- Pre-order page: one week. A payment link, a clear promise, a stated delivery date and a refund policy you will honour. Check how taking money in advance is treated where you live before you switch it on.
- Deposit for a service: two evenings. A proposal, a price, a deposit invoice. Most of the week goes on finding people to send it to.
- Paid pilot: one week to sell, then real delivery time. This is the only rung that eats weekends, and it is the only rung that pays you.
Notice the pattern. The strongest evidence is not the most expensive to test. A pre-order page costs about the same week as a landing page. The reason people choose the landing page is that it cannot reject them.
Start at the top and climb down
The usual advice is to work up the ladder. Do the opposite. Start with the strongest test you can plausibly run this month, and only fall back a rung when you genuinely cannot get to a buyer.
If you sell to businesses and you can reach a decision maker, ask for a paid pilot in the first conversation. If you sell to consumers and the thing is deliverable by hand, sell it by hand before you automate anything. The first customer is a validation instrument, not just revenue. If you have no idea what to charge in that first conversation, work out a defensible price before you knock, because a number you invent on the call will be the number you live with.
A hundred people on a waitlist and one person who paid you a deposit are not comparable numbers. The one who paid is the finding. The hundred are a hypothesis wearing a bigger font.
What to do when nobody climbs
Sometimes you run the test and everyone stops at the free rung. That is not failure. That is the cheapest possible answer to "is there demand for my product", delivered before you spent a year finding out the slow way.
Take the specific objection you heard most and treat it as the new brief. Wrong buyer, wrong problem, wrong price, wrong moment. Usually it is one of those four, and usually the interviews already told you which. Some of these are early signs the idea will fail, and catching them in week three is a good week's work.
If you want a fast read on where your idea currently sits, run it through the free five-minute Idea Validation Score and see which rungs you have actually cleared. If you would rather have someone run the top-rung test with you and go looking for a real paying buyer, that is what the First Customer Validation is for: the what it covers page explains the work, and how pricing works is a conversation rather than a printed number. The broader method sits in the guide to validating a business idea.
Frequently asked questions
How do I test demand if I have nothing to sell yet?
Sell the outcome, not the product. Write down exactly what the buyer gets, by when, for how much, then ask a real prospect to pre-order it or pay for a one-off pilot you deliver manually. If they say yes, you have a customer and a build spec at the same time. If they say no, the reason they give is the most valuable market research for a small business you will get all year.
How many pre-orders or deposits do I need before I start building?
Fewer than you think, and more than one. One buyer can be a friend, a fluke or a uniquely desperate case. Three to five unrelated buyers paying a similar price for a similar promise is usually enough signal to justify building, because the pattern across strangers is what matters, not the total.
Are surveys a waste of time for testing demand?
Surveys are fine for describing a market and terrible for predicting purchases. Use them to learn what tools people currently pay for, how often the problem occurs and who owns the budget. Never use a survey question about future intent as your reason to build, because answering costs nothing and people are consistently generous with hypothetical money.