How to Validate a Business Idea Before You Build It

Ask your friends whether they would use your app and most of them will say yes. Ask them for twenty pounds today and you find out what the yes was worth. That gap, between what people say and what they do, is the whole reason idea validation exists.

It is also why so many products die after launch instead of before it. CB Insights reviewed 431 venture-backed companies that shut down since 2023 and published the analysis in March 2026. Running out of money was the final cause in 70 percent of cases, but that is usually the moment of death rather than the disease. Further down the list sits the more useful number: 43 percent cited poor product-market fit. Those teams had boards and advisers. Solo founders with a spreadsheet have even fewer people around to say “this won’t sell”.

Think in evidence, not opinions

The most practical way to approach validation is to rank every signal you collect by what it cost the person who gave it. Praise costs nothing. Time costs a little. Putting their name or reputation behind you costs more. Money costs the most. The higher a signal sits, the more you can trust it.

1. Compliments"Great idea, I'd use that." Pleasant, nearly worthless on its own.
2. Stories about the pastSpecific recent incidents, what they tried, what it cost. Real data about the problem.
3. TimeA second meeting, a demo with their team, filling in a detailed form.
4. ReputationAn introduction to a colleague or their boss, a public testimonial for a beta.
5. MoneyA pre-order, a deposit, a paid pilot. The only rung that proves willingness to pay.

Adapted from the commitment logic in Rob Fitzpatrick's The Mom Test. The ranking is a judgement aid, not a measured scale.

Everything that follows is a way of climbing that ladder as cheaply as possible, and of noticing early when nobody is willing to take the next step.

Week one: decide what could kill the idea

Every idea rests on assumptions. Usually there are three kinds: the problem happens often enough to matter, someone will pay to have it solved, and you can reach those people for less than they pay you. Software adds a fourth that founders skip: customers must give up whatever they use now. Almost every problem already has a workaround, whether a spreadsheet, a shared inbox or a junior employee. Habit is your real competitor.

Turn each assumption into a sentence that a result could disprove. Take a founder in Leeds thinking about software for independent letting agents. “Letting agents need better tools” proves nothing either way. “Agents managing fewer than 200 properties spend more than five hours a week chasing contractors for repair updates” can be confirmed or destroyed in a dozen phone calls.

Then rank by two questions: how unsure are you, and how fatal would being wrong be? Test the most uncertain, most fatal assumption first. Write next to it the result that would prove it false, before you collect any data.

Weeks two to five: climb past compliments

The problem gets tested in conversations, and the standard reference is Rob Fitzpatrick’s The Mom Test. Talk about the other person’s life, not your idea. Ask about the last time the problem happened. What someone did last Tuesday is data. What they say they will do next quarter is politeness.

How many conversations? A 1993 study by Abbie Griffin and John Hauser found that 20 to 30 interviews surfaced roughly 90 percent of customer needs within a segment. Treat that as a planning figure. Ten calls tell you whether you have the right segment at all. By 25, you will be hearing the same stories.

End every call by asking for something from a higher rung: an introduction, a second meeting with a colleague, access to the messy spreadsheet they use today. People who decline every small ask rarely turn into customers. The full question list is in our guide to customer discovery interview questions.

Weeks four to eight: ask for money

Interviews establish the problem. Only an offer with a price shows willingness to pay. For consumer and self-serve products, the usual tool is a landing page that sells the product as if it existed, then explains honestly after the click. How to run one and read the numbers without fooling yourself is covered in our guide to the smoke test landing page.

In B2B you can often go straight to the top rung. If several interviewees describe the problem as genuinely expensive, propose a paid pilot with a start date and a price. Even a letter of intent with both is meaningful. “Sounds great, let me know when it’s ready” is not.

A back-of-envelope market check

Once the problem and the willingness to pay look plausible, check whether the numbers can carry a company. Skip analyst reports about “a global market worth billions” and count from the bottom: how many businesses have this exact problem, how many can you reach, and what will each pay per year?

The UK government’s business population estimates put the number of private sector businesses at about 5.69 million at the start of 2025, and roughly three quarters had no employees. If your product only makes sense for firms with staff, most of that headline figure is irrelevant before you start. Our walkthrough on how to calculate TAM, SAM and SOM shows the method end to end.

Pre-commit to a kill rule

Most founders skip this, and it is the step that saves the most money. Before your first call, write down the results that would make you stop, and give the list to someone who will hold you to it.

• Fewer than a third of the people you interview experienced the problem in the last three months.
• Nobody has spent money or real effort trying to solve it.
• Thirty conversations produce no pilot, pre-order or deposit.
• Acquiring one interested lead in your test already costs more than a customer would pay in year one.

Failing a rule does not always mean starting from scratch. Interviews often surface a neighbouring problem that is more frequent or more painful, and you can rewrite the assumption and go again. What you must not do is relax the rule afterwards because you have grown attached to the idea.

Where validation misleads

False positives happen. Twenty enthusiastic interviews do not guarantee twenty customers, and products that require a new habit often convince only once people use them.

False negatives happen too. Some ideas test badly because people cannot picture something genuinely new, and they succeed anyway. That is rare, and if you are betting on it, insist on at least one hard signal, such as a handful of users who paid.

When the evidence is strong, the next step is the smallest product that delivers the core result, and our breakdown of MVP development cost shows what that usually takes. If you would rather have an engineering partner than a cheque, look at how venture studios such as OnlineBuilders work with early founders.

This week: write down your three most dangerous assumptions, each with the result that would disprove it. Then book five conversations.

Step by step

1. List the risky assumptions Write down the two or three assumptions the business depends on, phrased so that a result could prove them wrong.
2. Interview people with the problem Hold 20 to 30 conversations and ask about specific recent incidents rather than opinions about your idea.
3. Test willingness to pay Put a priced offer in front of strangers before the product exists and count pre-orders, deposits or signed pilots.
4. Size the market bottom-up Estimate how many reachable customers exist and what they would pay per year.
5. Decide against your kill criteria Compare results with the thresholds you set beforehand, then continue, change the assumption, or stop.

Questions people ask

How long does idea validation take?

Four to eight weeks is a realistic first round if you can reach your target group. If it drags on much longer, the usual reason is that you cannot find the people. That tells you something uncomfortable about your future sales channel.

Is a survey enough to validate an idea?

No. Surveys record what people believe about themselves, not what they do. Use one after your interviews to estimate how common a pattern is, never as your only evidence.

Do I need a prototype to validate?

Not for the problem. For the solution, a clickable mock-up or a landing page helps, but only once you know the problem is real and painful.

What counts as a validated idea?

Something that costs the customer: a pre-order, a deposit, a signed pilot, or at least a meeting with the person who controls the budget. Compliments and newsletter sign-ups are weak signals on their own.