How to Validate a Business Idea Before You Quit Your Job

Inkdome began as a plan for a search engine to help people find the right tattoo artist. Its founders, one of the teams in a 2016 randomized trial in Milan, wrote the plan down as four hypotheses and set a minimum share of 50 customer interviews that had to support each one. Three passed. The fourth failed, and many interviewees described a different problem: not finding a tattooist, but judging one’s skill. The team rebuilt the idea as a platform where users seek advice from tattoo experts.¹

That is how to validate a business idea: treat it as a set of guesses that could be wrong, test the riskiest ones cheaply while your paycheck still arrives, and decide before each test what result would make you stop. The five steps:

1. Turn the idea into guesses that can fail
2. Ask strangers what they already do
3. Ask for a commitment, not a compliment
4. Run the cheapest test that could prove you wrong
5. Fix the pass mark and the quit point in advance

Whether to leave employment for a business at all is an earlier question, covered in the case for and against starting a business; this page assumes you have an idea to test.

Prefer to watch? The video covers the same ground as this article.

VideoHow to Validate a Business Idea Before You Quit Your JobWiserHours

How to validate a business idea the way a scientist tests a theory

Validating a business idea means testing the beliefs it rests on before you commit money and a career to it. The strongest evidence on how to do that comes from randomized trialsrandomized controlled trial: A study that assigns participants to the treatment or the comparison group at random, so the groups start out alike and a difference in what happens next can be put down to the treatment. Random assignment makes the two groups comparable; it does not make the people in the trial representative of anyone else.Full entry in the glossary of founder training in Italy and the UK, in which the method taught was the one scientists use: a theory, hypotheses drawn from it, and tests with a pass mark.

In four randomized trials run in Italy and the UK from 2016 to 2019, founders taught to test explicit hypotheses were more likely to drop an idea, and earned slightly more revenue on average, than founders given the same training without that method, Arnaldo Camuffo and colleagues reported in 2024.²

Why would a way of thinking change results? Most new business ideas eventually fail, the researchers reason, so the likelier mistake is a false positive: a weak idea that seems to be working. They argue that a founder with written hypotheses and a pass mark reads early signals more accurately, which makes them more willing to drop or reshape an idea that is not working.¹

The study

Moderate evidence

Hypothesis-trained founders dropped more ideas in four Italian and UK trials

Both groups took the same entrepreneurship course, covering customer interviews, surveys and experiments. Only the treated group was taught to turn its business model into a theory, derive hypotheses, test them and judge the results against them. Across the four trials, the method raised the chance that a firm dropped its idea during the study by about 10 percentage points, at a time when about a third of all the firms dropped theirs. Trained founders were more likely to change course exactly once and less likely to change more than twice. They also earned more revenue on average, an effect the authors call small because many firms had not yet earned anything.²

The researchers designed and supervised the training they tested, and the firms were young ventures in business-support programs, so the results speak to early decisions, not to long-term survival.² The first trial, in Milan, found more changes of course and more revenue among trained founders but no clear effect on dropping ideas, and few of its firms earned any revenue in the year it was followed.¹

Take an idea such as weekly meal boxes for night-shift nurses. Said one way, it is a hope: nurses would love this. Said another, it is a claim that can fail: “Nurses on night shifts will pay each week for meals delivered before a shift.” Only the second version can bring you unwelcome news in time to act on it. So write down what must hold for your business to work, and treat each line as something to test.

A five-step routine for evenings and weekends

The five steps below turn the trials’ method into a routine that fits around a job, each one cheap enough to run before you resign.

1. Write the guesses: each part of the idea as a hypothesis you could be wrong about
2. Test cheaply: the smallest test that could prove a guess wrong, such as asking for a pre-order
3. Set the bar first: the result that counts as a pass, written down before the test runs
4. Continue or change: in the trials reported in 2024, trained founders were more likely to change course exactly once and less likely to change more than twice
5. Stop: in the same trials, the method made founders more likely to drop an idea

1. Turn the idea into guesses that can fail

List what must be true for the business to work: who the customer is, what problem they have, whether they will pay, how you will reach them and what it will cost you to deliver. In the replication, trained founders started from a business model canvas, a standard template of a business’s parts, and were asked to explain why its parts would create value and to turn that explanation into testable hypotheses.² Put the guess that would sink the business if wrong at the top, and test it first.

2. Ask strangers what they already do

Talk to people in your target market whom you do not know, and ask about what they have already done. The interview guide given to trained founders in the Milan trial asked about the hardest part of the problem, the last time it happened, why it was hard and what the person had already tried.¹ If someone has never looked for a fix, treat their interest in yours with caution.

3. Ask for a commitment, not a compliment

Look for evidence that costs the customer something: a refundable pre-order, a deposit, a signed letter of intent or a paid pilot. A 2020 meta-analysis of 77 studies by Jonas Schmidt and Tammo Bijmolt found that consumers’ stated willingness to pay ran, on average, about a fifth above their real willingness to pay.³ A price someone pays is therefore a better guide than a price they name. Plan how you would refund people if the test fails, and never take money for something you could not deliver. If you are still employed, check your contract for limits on outside work, and your country’s tax rules, before charging anyone.

4. Run the cheapest test that could prove you wrong

Match the test to the question. A hand-run version of the service can show whether customers come back; a simple web page with a sign-up button can show whether strangers are interested; a rough prototype can show whether the thing works. For validation, spend only what the question needs.

Online, testing can be very cheap. In an observational study of roughly 35,000 tech start-ups worldwide in the Crunchbase database, tracked weekly from 2015 to 2019, Rembrand Koning, Sharique Hasan and Aaron Chatterji linked adopting A/B testing tools to faster growth in website visits and more new products, and to more very good and very bad weeks. Firms chose whether to adopt, which the authors say could inflate the estimates.⁴

5. Fix the pass mark and the quit point in advance

Before each test, write three things: the result that counts as a pass, the date you will judge it by, and your next move if it falls short. Inkdome’s rule named its interview count and the pass mark each hypothesis had to reach.¹ Give a copy to a person who gains nothing either way, and cap the time and money each test may take; how much cash to hold before leaving a job deserves its own calculation, once the idea has passed.

The replication’s authors suggest that a method like this may curb overcommitment to ideas that are not working.² The pull it counters is the sunk-cost effectsunk-cost effect: The pull to keep putting time, money or effort into something because of what has already been spent on it, rather than because of what it will bring from now on.Full entry in the glossary: the more months you have put in, the harder it is to walk away, and that is when a rule you wrote before the first test is worth most.

When a test misses its mark, the rule leaves three honest options: keep going with a changed offer, change course to a different customer or problem, or stop.

Before you trust your idea

• Written list of what must be true, riskiest guess first
• Conversations booked with strangers in the target market
• Questions about their past, not about my idea
• A test that asks for money or another real commitment
• Pass mark, deadline and next step written before the test
• Rule shared with a neutral person

Download PDF ↓

Get this checklist as a one-page PDF. We'll email you the download link.

Your emailAlso send me the weekly newsletter (launching soon)Email me the PDF

We use your address to send this link and, only if you tick the box, the newsletter. Privacy policy.

Why compliments are weak evidence of demand

Encouragement from people you know is weak evidence of demand because agreeing costs them nothing. In the Milan trial, founders left to their own methods often posted questionnaires to their own social-media contacts and asked whether people would use the service; the researchers point out that friends do not want to disappoint, and that saying yes to a service you will never have to pay for is free.¹

The gap between words and wallets is also larger for some offers than others. The Schmidt and Bijmolt meta-analysis found the overstatement was wider for specialty products and, less clearly, for higher-priced ones,³ so if your idea is a premium or niche offer, expect warm words to flatter it more.

Weak demand is also a common story in shutdowns. In its 2026 review of 431 venture-backed start-ups (countries not given) that publicly shut down since 2023, CB Insights, a company that sells data on private firms, found that 43 percent of those with identifiable reasons cited poor product-market fit; running out of money, cited more often still, is usually the last cause rather than the root one, the company says. These are public accounts from a self-selectedself-selection: When the people in a study chose to take part, or chose the thing being studied, rather than being picked or assigned. Volunteers and choosers can differ from everyone else, so the result may not apply more widely or may reflect who chose rather than what they chose.Full entry in the glossary group, not a random sample of businesses, so read them as a warning, not a rate.⁵

Many would-be founders skip even the simplest checks. In a nationally representativenationally representative: Describes a sample drawn by random selection from a country's whole population and weighted so its results stand for that population. The label describes how people were picked, not how many of them agreed to take part.Full entry in the glossary survey of Americans, first published in 2019, Victor Bennett and Aaron Chatterji found that fewer than half of people who had considered starting a business took even the cheapest steps, such as searching online for competitors or talking the idea over with a friend.⁶

Count the strangers who book a paid trial, and ask the questions on the right.

Further reading

• 

  The Mom Test

  by Rob Fitzpatrick

  A short guide to asking customers about what they already do, not whether they like your idea, as step 2 of this article advises.

• 

  The Lean Startup

  by Eric Ries

  Ries's method of treating a business idea as experiments with a pass mark, close to the approach tested in the trials here.

As an Amazon Associate WiserHours earns from qualifying purchases.

Where the evidence for testing runs thin

The randomized evidence is encouraging but narrow. Before the replication, the Milan trial was the only published test of this method, its authors note, and all four trials come from one research team working mostly with early-stage firms in training programs.² A 2025 review of research on entrepreneurial experimentation from 1985 to 2024 describes the field as fragmented and conceptually unclear.⁷

Nothing here shows that testing makes a business succeed, or tells you how long to test. A local café and a phone app need very different tests, and the trials cannot say which cheap test suits which business. They back a modest, practical point: a clear method for reading early evidence made founders more likely to drop ideas.

Who to ask before leaving your job

Testing costs evenings; resigning costs a salary, and whether you can absorb that depends on savings, debts and who relies on your income. A parent with a mortgage faces a different sum from someone with savings and no dependants, so line up outside help in order of urgency.

• If repayments are already a struggle, free debt advice is worth getting before you put money into the idea, from a public or charitable debt adviser in your country. Two examples: the CFPB, a US government agency, explains that the typical credit counseling organization is a non-profit giving free or low-cost advice, with fees possible for some services;⁸ the UK government’s debt page names MoneyHelper as a source of information on free debt advisers.⁹
• If resigning hinges on savings, pension, tax or debts, a qualified accountant or regulated financial adviser can work through your personal numbers; ask up front how they are paid.
• For help with the business itself while you test, check whether a public small-business agency in your country offers advice. For example, the US Small Business Administration reports that it and partner organizations offer counseling and training in your area at low or no cost;¹⁰ the UK government lists free business advice by phone, email or webchat, depending on which of the four UK nations you are in.¹¹

The bottom line

Give your idea a fair chance to fail cheaply. Write the guesses down, test the riskiest one while your job still pays the bills, and judge the result against a bar you set before the test. Count money or another real commitment as evidence, and treat praise as noise.