What to Validate at Each Stage

Validation asks a different question at each stage. Before building, whether the problem exists and someone already pays badly to solve it. Before charging, whether anyone will pay you specifically. During the MVP, whether people use it rather than sign up for it. Before scaling, whether acquisition works at a cost the business can afford. Asking them out of order is how six months disappear.

The mistake that costs six months

The common failure is not skipping validation. It is doing it once, at the start, and treating the answer as permanent.

Someone has an idea, asks fifteen people whether it sounds useful, hears yes fourteen times, and treats that as a green light for everything that follows. Six months later the product exists, the same fourteen people are polite about it, and nobody has paid.

Nothing went wrong with the test. The test answered a question, and it was the wrong question to have asked only once, because “does this sound useful” and “will you pay for this” and “will you keep using it” and “can I acquire people like you profitably” are four unrelated questions with four different answers.

Each belongs to a stage. Each has a cheap test. And each has a result that should make you stop, which is the part people skip.

Stage 1, before building: does the problem exist?

The question: is somebody already spending time or money on this problem in a way that is clearly bad?

What not to do: ask people if they would use your idea. Interest is free and everyone is polite.

The tests that work. Find the workaround. Look for the spreadsheet, the part-time hire, the three tools taped together, the agency doing it manually. Read complaints in the customer’s own words in forums and reviews. Look at job listings for the task you would automate, because a company hiring for it is a company paying for it.

What should stop you: you cannot find anyone doing it badly, only people agreeing it would be nice.

Stage 2, before charging: will they pay you?

The question: not whether the problem is worth money in general, but whether this person will hand money to you specifically.

What not to do: confuse enthusiasm for commitment. Verbal agreement costs nothing.

The tests that work. Presell, with a deposit or a paid pilot. Offer to do it manually for a fee before anything is built, which also teaches you the workflow. Put up a price and see who asks to buy rather than who asks for a demo. Letters of intent from businesses are weak; small payments are strong.

What should stop you: repeated enthusiasm with no payment. Five yeses and zero deposits is a no.

Stage 3, during the MVP: do they use it?

The question: do the people who signed up actually use it a second time?

What not to do: measure signups, waitlists or downloads. Those measure your marketing, not your product.

The tests that work. Track the second use and the fourth. Watch what people do with the part you thought was secondary. Ask the ones who stopped, which is uncomfortable and the most useful conversation available. Look at whether they would be annoyed if it disappeared.

What should stop you: people sign up, use it once and go quiet, and when you ask, they cannot articulate what it replaced.

Stage 4, before scaling: does acquisition work?

The question: can you get customers repeatably, at a cost that leaves a business?

What not to do: scale on the strength of customers who came from your own network. Friends, former colleagues and your existing audience are not a channel.

The tests that work. Find one channel where a stranger becomes a customer. Track what it costs, including your time. Compare that to what a customer is worth over a realistic lifetime, not a hopeful one. Repeat it twice before you believe it.

What should stop you: every customer traces back to somebody you already knew, or acquisition costs more than the customer will ever be worth.

The four stages at a glance

Stage	Question	Good evidence	Stop if
Before building	Does the problem exist?	A workaround someone pays for	Only polite agreement
Before charging	Will they pay you?	Deposits, paid pilots, prepayment	Five yeses, zero payments
MVP	Do they use it?	Second and fourth use	One use, then silence
Before scaling	Does acquisition work?	A stranger acquired repeatably	Everyone came from your network

Why the order matters

Each stage is cheaper than the one after it, so running them in sequence means you spend the least money on the ideas most likely to fail.

Running them out of order inverts that. Building before you know the problem exists means the most expensive stage happens first. Scaling before you know people use it means paying to acquire customers who will leave.

The purpose of validation is not to prove yourself right. It is to find out you are wrong while it is still cheap, and the sequence is what keeps it cheap.

FAQ

How do you validate a business idea? In stages. Confirm the problem exists by finding what people pay for as a workaround, confirm they will pay you by taking money before you build, confirm they use it by tracking repeat use rather than signups, and confirm acquisition works by winning a customer who is a stranger.

How many people should you talk to? Fewer than most advice suggests, and more specifically. Five conversations with people who have the problem beat fifty with people who find it interesting.

Is a waiting list good validation? It is weak. Signing up costs nothing. A deposit, a paid pilot or a prepayment is a different kind of signal, and it is the one worth collecting.

How long should validation take? Days to weeks per stage, not months. If a validation step is taking months, it is usually because it has quietly become building.

What if the idea fails validation? Then it failed cheaply, which is the entire point. Most ideas should end at stage one or two, and that outcome saves the months you would otherwise have spent.