How to validate a startup idea before you build it

What validation actually means

Validation is not asking friends whether your idea sounds good. It is not a survey where people say they would probably use it. Both of those reliably produce false positives, because people are agreeable and because a hypothetical yes costs nothing.

Useful validation looks for evidence that already exists in the world, independent of you: people complaining about the problem in public, money already changing hands to solve it badly, competitors who exist and are growing, or a segment that every existing product visibly ignores.

The bar to clear is not "does this sound like a good idea." It is "is there evidence this problem is painful enough that someone already paid to make it go away."

Step 1: Write the problem down without mentioning your solution

Most idea descriptions are solutions in disguise. "An app that helps freelancers track invoices" is a solution. The problem underneath it is something like "freelancers lose money because they forget which invoices went unpaid, and chasing them feels rude."

This matters for a practical reason, not a philosophical one: you cannot search for competitors properly until you can describe the problem in the customer's language. Search for your solution category and you will only find companies that describe themselves the way you do, missing everyone solving the same pain differently.

Step 2: Name the specific person who has it

"Small businesses" is not a customer. "A solo design freelancer in their first two years, billing five to ten clients a month, with no accountant" is a customer. The narrower version is testable and the broad one is not.

This is also where most competitive analysis goes wrong. Two products can look identical and serve completely different buyers, and that mismatch is usually where the real opening is.

Step 3: Find who is already live

Search for the problem in the words your customer would use, not your category name. Then widen: look for indirect competitors and manual workarounds. A spreadsheet, a WhatsApp group, or a virtual assistant is frequently the real incumbent, and it is the hardest one to displace because it is free and already trusted.

For each competitor you find, write down three things:

• What they actually do, in one sentence, from their own site rather than your assumption.
• Who they are built for, in one phrase.
• What they visibly do not do.

Finding nothing is a warning, not a win. It usually means you searched with category wording, or the market was tried and abandoned because the pain was not worth paying to fix.

Step 4: Decide whether a real gap exists

Look at the who-it-is-for column of your competitor list. If every entry points at the same buyer, ask who else has this problem and is absent from the list. That absence is your opening.

Then apply the honesty test. If the thing you would add is a feature an incumbent could ship in a sprint, it is not a wedge, it is a roadmap item on someone else's roadmap. A real wedge is usually a different way of working, a different buyer, or a different business model, not a better version of the same screen.

Step 5: Check whether the pain has a price

A problem people acknowledge is not the same as a problem people fund. Look for evidence of money already moving: paid competitors with real pricing pages, agencies or freelancers charging to do it manually, or an expensive internal workaround.

If nobody anywhere is paying anything to solve this today, that is the single strongest signal to stop. It is also the cheapest one to find, which is why it belongs before you build rather than after.

Step 6: Talk to five people who have the problem

Desk research tells you whether the space is open. Conversations tell you whether the pain is real. Ask about the last time the problem happened, what they did about it, and what it cost them. Past behaviour is evidence; future intentions are not.

Do not pitch. The moment you describe your solution, the conversation becomes a politeness exercise and the data is gone.

Step 7: Run the smallest test that could fail

A good test has a real cost to the other person: a pre-order, a deposit, a signed pilot, a scheduled call. A landing page with an email box is weak evidence because an email address is nearly free to give.

Design the test so a negative result is unambiguous. If you cannot describe in advance what result would make you abandon the idea, the test is decoration.

How long this should take

Steps 1 to 5 are desk research and can be done in a couple of focused days. Steps 6 and 7 take a week or two depending on how reachable your customer is. That is the honest budget: a fortnight, against the several months a wrong build costs.

Founder Evolution AI compresses steps 1 to 5. You describe the idea in plain language, it runs live competitor research, judges whether a genuine gap remains, rewrites the idea into a sharper version when one does, and returns a Go, Pivot, Saturated, or Stop verdict with its reasoning and sources. It does not replace steps 6 and 7, and does not claim to. It tells you whether those conversations are worth setting up.