In short: To validate a business idea, gather evidence that specific people will pay to solve a problem before you build the full solution. Talk to potential buyers about what they have already done, test demand with a simple page or manual service, and try to collect money. Compliments and survey answers count for very little.
Enthusiasm is cheap. Friends will say an idea sounds great because they like you, and asking “Would you use this?” invites a polite yes. What you are looking for is a commitment that costs someone something: time, a referral, a deposit or an actual payment.
There is good reason to bother. When CB Insights studied 101 post-mortem essays written by founders of failed startups, the most common explanation, cited in 42% of cases, was no market need. Running out of cash came second at 29%. The study dates from 2014, relies on self-reported reasons and covers venture-style startups more than small side projects. Still, the lesson travels well: a lot of failures involve building something few people wanted.
Write the idea as a claim you can test
Vague ideas can’t fail, so they can’t be tested either. State yours like this:
[Specific group] currently deals with [specific problem] by [current solution], and they would pay [price] for [your solution] because [reason].
For example: “Independent physical therapists currently chase unpaid insurance claims by hand, and would pay $49 a month for a tool that flags them, because it recovers more than it costs.” Each part can now be checked, and each part could be wrong.
Next, list what must be true for the business to work. Does the problem occur often? Do buyers control a budget? Can you reach them at a reasonable cost? Test the riskiest assumption first, which is usually whether anyone cares enough to pay.
Talk to buyers about the past
Interview 10 to 15 people from your target group. Don’t pitch. Ask what has happened, since behavior predicts more than opinions do.
| Weak question | Better question |
|---|---|
| Would you pay for a tool like this? | When did this last happen, and what did you do? |
| Do you think this is a good idea? | What have you already tried to fix it? |
| How much would you pay? | What does the problem cost you now? What do you pay for your current solution? |
| Is this a big problem for you? | What did you stop doing to deal with this? |
Listen for evidence of effort. Someone who has built a messy spreadsheet, hired a temp or paid for a poor tool has a live problem. Someone who says “Yes, that’s annoying” but has never acted on it probably won’t buy. Note the exact words people use too, since they make the best marketing copy.
Run a test that resembles the business
Pick the cheapest test that can actually produce a clear answer.
| Test | How it works | What it tells you | Rough cost |
|---|---|---|---|
| Landing page | Describe the offer with a sign-up or “buy” button | Whether the message attracts interest | Low |
| Small ad campaign | Send targeted traffic to the page | Cost of reaching buyers | A modest ad budget |
| Pre-order | Take payment or a deposit with a refund promise | Willingness to pay | Low |
| Manual service | Deliver the result by hand for a few customers | Whether the result has value | Your time |
| Explainer video or demo | Show how it would work | Interest in the concept | Low to medium |
Two well-known stories show the approach. Zappos reportedly began in 1999 with Nick Swinmurn posting photos of shoes from local stores online, then buying a pair from the store when an order came in. There was no warehouse, just a test of whether people would buy shoes on the internet. Dropbox made a short explainer video before its product was widely available, and its waiting list reportedly jumped from 5,000 to 75,000 sign-ups overnight. Retellings of such stories tend to be tidy, but the logic stands: they learned what they needed before they built the expensive part.
Doing things manually at first has a further benefit. If you plan an automated service, do the work yourself for your first three customers. You’ll learn what they really want, and what you can safely leave out.
Weigh the evidence
Not all signals are equal. Roughly in order of strength:
- Repeat payment from strangers
- A first payment from strangers
- A deposit or a signed commitment
- An introduction to a colleague who has the same problem
- A scheduled follow-up meeting
- An email sign-up
- A like or a comment
- “Great idea, I’d definitely buy that”
Friends and family don’t count as strangers. Be careful about sign-ups as well, because leaving an email address costs almost nothing. A pre-sale, even for a small sum, carries much more information. If you take money before a product exists, say so clearly, give a delivery date and refund anyone who asks. In the US, advertising a physical product with a shipping date also brings FTC obligations, which we describe in how to start dropshipping.
Decide beforehand what would stop you
Founders tend to keep moving the goalposts. Set your pass mark before the test starts, and write it down. For instance: “If 20 outreach conversations produce fewer than three paid pre-orders at $99, I will change the offer or stop.”
The numbers are your judgment, since there is no universal benchmark. Base them on what the business needs. If you need 100 customers at $50 a month to make the effort worthwhile, and it took 40 conversations to get a single one, you can work out whether that sales effort is realistic.
A failed test can lead to several outcomes:
- Different customer. The problem is real, but a different group feels it more.
- Different offer. Buyers want a service instead of software, or a template instead of a course.
- Different price. Your cost may be fine, but the framing is wrong.
- Stop. This isn’t a defeat. A month spent learning an idea doesn’t work costs less than a year spent building it.
Common traps
- Mistaking competition for a bad sign. Competitors usually show that people pay in this market. Look for an unhappy group or a neglected niche.
- Using surveys alone. They measure what people say. Use them to shape questions, then test behavior.
- Building “just a quick version” first. Technical people especially tend to start coding to avoid sales conversations, which are the part they need most.
- Stopping at validation. Evidence of interest doesn’t prove a business works. You still need to test the cost of reaching customers and the margin. Our guide to digital products looks at how fees and pricing change the arithmetic, and how to sell digital products takes you through a first launch.
Validation doesn’t remove risk. It moves the risky decision to a point where being wrong costs little. For how this step fits in with everything else, see how to start an online business.
Frequently asked questions
Won’t someone steal my idea if I talk about it?
It’s unlikely, since others are busy with their own work and execution is the hard part. The danger of building in secret is usually much bigger. If the idea involves a true trade secret or something patentable, take legal advice before disclosing those details, while still testing the problem and demand.
How many people do I need to talk to?
There’s no fixed number, but 10 to 15 conversations within one tightly defined group often reveal repeating patterns. Stop when you keep hearing the same things. If the answers are all over the place, your target group may be too broad, and narrowing it is usually more useful than adding interviews.
Can I validate an idea using AI tools or search data?
They help with preparation. Keyword data can show that people look for a solution, and AI can help you draft questions or scan competitors. Neither can tell you whether a real person will pay you. Treat them as a way to form better hypotheses, then test those with actual buyers.
