How to Start
a Business

A practical guide

How to validate a business idea

Validate an idea by finding out whether people with the problem will give you time, then money, then a paid pilot, and treat anything less as encouragement.

Published November 3, 2026 · Luke F. Walton

You validate an idea by asking people who have the problem to give you something that costs them. First their time, then a deposit or pre-order, then a paid pilot. Praise, likes and “let me know when it’s ready” do not count.

The test can be small. A handful of honest conversations and one request for money will tell you more than a month of planning.

What counts as evidence?

Evidence is something a person did that cost them time, money, or effort. Encouragement is something they said that cost nothing.

The SBA frames market research as a way to reduce risk and confirm your idea before you spend. Its list of questions is a good starting checklist. Is there demand? How many potential customers are there, and where are they? What do local incomes look like? How many alternatives exist, and what do customers pay for them? You are answering the same questions, with your own legwork in place of a report.

Say you want to start a mobile bike repair service. “Great idea, I would use that” is encouragement. A neighbor who books Saturday at ten and pays a deposit has given you evidence.

What is the ladder of commitment?

It runs from cheap to costly, and each step proves more than the last.

  1. A conversation. Someone gives you twenty minutes and talks about the problem. This proves the problem is real enough to discuss.
  2. A pre-order or deposit. Someone pays before the thing exists. This proves they want your solution, at your price.
  3. A paid pilot. Someone pays for a real, limited version of the work. This proves you can deliver and they value the result.

Climb as far as the idea needs. A laptop consulting practice can often reach a paid pilot with one client in a month. A small shop with a lease may only reach a pre-order or a weekend pop-up before you sign anything. Stopping at a conversation is fine for a day or two. Staying there is a way of not asking.

How do you design a test you can run this month?

Pick one question, one group of people, one action, and a ceiling on spending. Then write all four down.

For the bike repair service: “Will commuters within two miles pay $35 for a flat fix at their door? I will talk to ten and try to book five visits. I will spend no more than $75 on a pump, patches and a printed card.”

The budget matters because it forces a small test. Pick an amount you would not miss if the idea fails. If you are unsure what a first test might cost, list the items in the startup cost calculator and take only the ones the test needs. Leave the rest for after you have an answer.

Set a date. A test with no end becomes a hobby.

How many conversations are enough, and what do you ask?

Aim for about ten to fifteen conversations with people who fit your described customer, and stop sooner if the answers are clearly one-sided. That is a rule of thumb, not a researched threshold. What matters is that you hear the same things repeat, or clearly fail to.

Do not ask “Would you buy this?” People guess about future behavior and lean toward being polite. Ask about the past, where answers are facts:

For the bike repair idea: “When did your bike last break? What did you do? What did it cost? Was getting it to the shop a hassle?” If someone has had the problem recently, tried to fix it, and spent money, you are talking to a real customer. If they cannot remember the last time, the problem is small.

Take notes in their words. Save the pitch for the end, and then make an offer with a price.

What do the results mean?

You will get one of three outcomes. Treat each differently.

A clear no. Most people do not have the problem, or they have it and will not pay. This is a good result. You spent a few weeks and a small budget, not your savings. Go back to your list of ideas.

A maybe. People like it, but nobody commits. Look for a reason before you decide. Maybe the price is wrong, the timing is wrong, or you are talking to the wrong people. Change one thing and test again. Do this once or twice, not forever. If nothing you change produces a commitment, treat it as a no.

A yes. A few people paid a deposit, booked a time, or signed up for a pilot, and they described the problem in their own words before you pitched. That is enough to start small. It does not mean the business will work at scale. It means you are no longer guessing about the first customers.

A simple check with example numbers: you speak to 12 people, 4 have had the problem recently, and 2 pay a deposit. That is 2 of 12, about 17 percent, and 2 of the 4 who really have the problem. The second ratio tells you more. It suggests you reached the wrong people for half of those conversations, which is something to fix in the next round.

How do people fool themselves?

Three ways show up again and again.

Friends and family being kind. Someone who loves you will say it sounds great. Talk to people who do not owe you anything, and weigh what they do over what they say. If you must start with friends, ask them to introduce you to strangers who have the problem.

Counting interest as demand. Likes, signups for a free list, and “keep me posted” feel like progress. They cost nothing. Count only commitments that cost something, however small.

Testing the pitch instead of the product. A good landing page or a persuasive talk shows you can sell a story. It does not show people want the thing. If your test measures how well you describe it, change the test so that someone has to pay, book, or hand over something real.

There is a fourth, quieter trap: hearing only the yeses. Write down every no, with the reason, as carefully as the yeses.

When do you stop testing and start?

Stop when another round would not change what you do. If you have a few people who paid, you know the price they accepted, and you know what the work takes, begin delivering. Further testing becomes procrastination.

Start small. Deliver to the people who committed, and let their experience tell you what to build next. Then record what you learned in your plan, marking what is now evidence and what is still a guess.

If the test failed, you have also learned something worth money. Go back to how to find a business idea and take the next candidate through the same steps.

Where to go next

Once you have a yes, read how to get your first customer to turn commitments into delivered work. When you want to organize what you learned into a full document, how to write a business plan covers the structure.

Common questions

Can I validate an idea without spending anything?

Often yes. Conversations are free, and you can ask for a pre-order before you buy materials. Spending a small, fixed amount can speed things up, but it is not required to learn whether people will commit.

Is a survey enough?

A short survey can help you find people to talk to, and the SBA lists written surveys among its research methods. But a survey records opinions. Use it to find candidates for real conversations, then ask for a commitment.

What if people say they would pay but then do not?

That gap is the finding. Ask what changed, and whether the price, timing or trust was the issue. If you cannot close it with one adjustment, count the original “yes” as encouragement.

What if my idea is for a market where nobody I know is a customer?

Find the places those customers already gather, online or in person, and ask for short conversations. Offer a fair trade for their time. It takes longer, which is a useful signal about how hard reaching them will be later.

Should I tell people my idea is confidential?

You rarely need to. Ideas are common, and delivering well is the hard part. Asking people to sign something before they talk to you usually slows your learning more than it protects you.

Put this in your plan

Read next