How to test a startup idea cheaply (before you spend a penny building)
Quick answer: You can test most startup ideas for almost no money. The real cost is honesty and a few days, not cash. Talk to ten people who have the problem, ask what they already do about it instead of pitching your solution, then ask for one thing that costs them something (a deposit, a pre-order, a real slot in their diary). Rob Fitzpatrick's The Mom Test sums up the cheapest instrument you own: everyone will lie to you politely, so measure what people do, not what they say. Building is the expensive way to learn what a conversation could have told you for free.
You have an idea and two hundred pounds.
The instinct is to spend it. Buy the domain, pay for a logo, take out a no-code subscription, maybe run a small ad. It feels like progress because money left your account.
Put the card away. You do not have a spending problem yet. You have a knowing problem, and money is a bad way to solve it.
Cheap does not mean weak
There is a myth that a cheap test gives you a cheap answer. That expensive validation, a real build, a real launch, a real budget, is the only thing that tells you the truth.
It is backwards. The most expensive test, building the whole thing, usually gives you the slowest and most confusing answer, because by the time it is live you have too much invested to read the result honestly.
Cheap tests are not the compromise. They are the point. You want to find out you are wrong before being wrong has cost you six months.
The cheapest test is a conversation, done right
Talking to people is free. Getting a useful answer out of them is the hard part, and it is where almost everyone goes wrong.
The trap is that people lie to you. Not out of malice. They lie because they like you, they want to be encouraging, and telling you your idea is flawed feels rude. So you pitch, they smile, they say it sounds great, and you walk out with a dangerous little dose of false confidence.
Rob Fitzpatrick wrote a whole short book about escaping this, called The Mom Test, and the title is the idea: ask questions in a way that even your mum could not lie to you, because you never mention your idea at all. Talk about their life. Ask about specifics in the past, not hopes about the future. Listen more than you talk.
"Would you use an app that did X" is worthless. "Tell me about the last time you dealt with X" is gold, because it is a fact, not a compliment.
That single shift costs nothing and changes everything.
Ask for something that costs them
A conversation tells you whether the problem is real. It does not tell you whether people will act. For that, you have to ask for a commitment that costs the other person something.
Money is the clearest currency, but it is not the only one. Time is a currency. Reputation is a currency, as in "would you introduce me to two colleagues who have this problem". A real slot in the calendar is a currency.
The test is simple. If the yes is free, it is worth nothing. If the yes costs them something and they still say it, you have real data. I have written more about turning that into a demand signal in how to test willingness to pay.
Applause is free. Commitment is data. That line is the whole discipline in five words.
The near-zero-cost ladder
Here is the order I would run tests in, cheapest first. Climb a rung only when the one below gives you a signal worth the next bit of effort.
Ten honest conversations. Free. Follow the rule above. If you cannot even find ten people who have the problem, that is your answer, and it cost you nothing.
A one-page landing page with a real call to action. A few pounds and an afternoon. Describe the thing as if it exists, and put a real button on it: pre-order, book a call, join a paid beta. A "notify me" box is weak. A button that asks for a commitment is strong.
A fake-door or pre-sell. Same page, but the button tries to take money or a deposit. Be honest at checkout that it is early access. What you are buying, cheaply, is the difference between "I would pay" and someone actually reaching for their wallet.
Concierge. Deliver the outcome for one person by hand, with no product at all, doing the work manually behind the scenes. It does not scale, and it is not supposed to. It tells you whether the result is even worth wanting before you automate it.
None of that is a build. All of it is learning.
Why people still fool themselves
Even cheap, well-run tests get misread, because the founder wants a particular answer. You hear the warm yes and file the shrug that came with it under "close enough".
That is a known failure, and it has its own post: why people say they love your idea but do not pay. The short version is that politeness is not demand, and the only cure is to keep asking for the costly yes rather than the free one.
Where a second opinion helps
The cheapest tests still run through the most biased judge available, which is you. When it is your idea, you round every ambiguous signal up.
That is the gap we are building Foxy to fill. It is an AI co-founder whose job is not to be nice about your idea but to read your customer conversations and tell you where the evidence is thin and which assumption you are quietly avoiding. If you want to run your idea through that before you spend anything building, start here.
So before you buy the domain, ask yourself one thing. What is the cheapest test that could prove this idea wrong this week, and are you actually willing to run it?
