Skip to main content

Validating an Idea Before You Build It

Zero to One · 10 min read ·

Cheap, honest ways to test whether anyone wants your product before you spend months building it: interviews, fake doors, pre-orders and manual versions.

Illustration: A risograph two-colour collage of a paper door with a hand-drawn handle labelled Coming soon, with torn-paper footprints leading up to it

The cheapest mistake in the world is building the wrong thing. It is also the most common. A founder spends four months making an app, launches it with a flourish, and discovers that the people she imagined using it have a perfectly good spreadsheet and no wish to change.

Validation is the habit of finding that out first. It does not need a budget, a team or special tools. It needs curiosity, a willingness to be told no and a few honest tests. This guide lays out the tests, from lightest to heaviest.

What you are trying to learn

Before running any test, write down the three things you actually need to know.

  1. Does the problem exist? Do real people have it, and does it bother them enough to act?
  2. Is my solution one they would use? Not "is it clever", but "would it fit into their life?"
  3. Would they pay, or give something of value? Money, time, data or effort.

Every test below answers one or more of those. If a test does not, skip it.

The lean startup approach describes the same idea as validated learning: treating your early work as experiments that test assumptions about customers, instead of building on faith. You are not trying to prove you are right. You are trying to find out what is true.

Test one: conversations

Start by listening. Find people who might have the problem and ask about their lives, not your idea.

Good questions:

  • "Tell me about the last time you did X."
  • "What was the hardest part?"
  • "What did you try to fix it?"
  • "What did that cost you, in time or money?"
  • "What would happen if you did nothing?"

Bad questions:

  • "Would you use an app that did this?" Almost everyone says yes to be polite.
  • "Do you think my idea is good?"
  • "How much would you pay?" Without a real offer, answers are fiction.

Take notes in their words. Look for patterns across conversations, and for stories where the person has already spent effort or money on the problem. Those are signs of real pain.

Ten conversations is a fair start. If the first five all tell the same story, you are probably onto something. If every story is different, the problem may be too vague.

Test two: look at what people already do

People's actions beat their opinions. Look for evidence of existing behaviour.

  • Are people asking about this problem in forums or groups? How often, and how urgently?
  • Do they share workarounds, templates or hacks?
  • Are there existing products, even clumsy ones, and do people complain about them?
  • Are people paying for related things?

Existing competitors are usually a good sign. They show demand. Your job is to find what they do poorly and for whom.

Test three: the manual version

Before you automate anything, do the job by hand for a few people. If your idea is a tool that turns receipts into tidy expense reports, offer to do it manually for three people for a week. You will learn what they actually want, which steps are hard and whether they come back.

This is sometimes called a concierge test, and it tells you more than a prototype would. It also often yields your first paying customer.

Test four: the fake door, honestly done

A fake door is a page that describes a product and invites people to sign up, before the product exists. It is a way to see whether people take action.

It can be done honestly, and it must be. Say clearly that the product is not available yet. Offer something fair in exchange for interest: a place on a waiting list, early access, a say in what gets built. Do not take money for something that does not exist unless you are clear about what happens and when.

Some practical points:

  • Be specific on the page. Say what the product does and who it is for in plain language.
  • Count actions, not views. How many people leave an email address, out of how many who arrive?
  • Tag your links so you know which channels brought which sign-ups.
  • Follow up. Email the people who joined and ask what made them sign up. Replies are gold.
  • Respect data rules. If you collect email addresses, be clear about what you will use them for, and only email people in ways they agreed to. UK guidance on direct marketing and electronic communications explains the consent rules that apply.
  • Do not mislead. Advertising rules in many places say marketers must not give a false impression. A page that implies a finished product is a misleading one.

A low sign-up rate does not always mean the idea is bad. It might mean the page is unclear. Try changing one thing, such as the headline, before you decide.

Test five: pre-orders and deposits

The strongest evidence is money. If a few people will pay before it exists, that is real. But it carries responsibility.

  • Be honest about what is being sold, when and what happens if you do not deliver.
  • Check the consumer rules that apply to your sales. They differ by country.
  • Keep it small and personal at first: ask three to five people you have spoken to.
  • Never take money you cannot return if you fail to deliver, and be clear about the terms before taking it.

Even without payment, other costly signals count: someone booking time for a trial, giving you access to their data or introducing you to a colleague.

Rank the evidence

Not all evidence is equal. From weakest to strongest:

  1. Compliments and "great idea".
  2. Saying they would use it.
  3. Joining a waiting list.
  4. Giving feedback in detail.
  5. Taking a trial or manual service.
  6. Paying or committing time and money.

Keep a note of how many people reach each rung. A product with ten compliments and no pre-orders is in a different position from one with three pre-orders and no compliments.

Beware of false positives

Some things feel like validation and are not.

  • Friends and family who want to be kind.
  • A spike from one post that shows curiosity, not need.
  • Sign-ups from people who were promised something unrelated.
  • Asking leading questions that guide people to the answer you want.
  • Only talking to people who already like you.

To guard against these, talk to some strangers, ask neutral questions and look for behaviour.

Beware of false negatives too

A failed test does not always mean the idea is wrong.

  • Wrong audience. The problem may be real for a different group.
  • Wrong message. People may need a different explanation.
  • Wrong channel. Your audience may not be where you looked.
  • Wrong timing. Some problems are felt at certain times of year.

Before you abandon an idea, ask which of these could explain the result, and run one more test.

A one-week validation sprint

Here is a shape for a quick sprint.

  • Day one: write your three questions and the person you imagine.
  • Days two to four: talk to five people. Take notes.
  • Day five: summarise patterns in a page. What did people say about the problem? What have they tried?
  • Day six: build a one-page description and share it with the same people. Count who responds.
  • Day seven: decide. Continue, change or stop.

Write the decision down, with your reasons. In a month you will want to remember why.

When the answer is no

Stopping early is a success, not a failure. You saved time, and you learned something real. Most good ideas come from a bad one that was examined honestly. Write down what you learned about the people you spoke to. It often points straight to the next idea.

On this site

If you pass validation and are ready to show the product, the submit page starts a listing. The search and categories pages show what others have made in your space, which is a quick way to check for existing products, and the launches page shows how small projects introduce themselves.

An example from start to finish

A founder wants to build an app that helps allotment holders swap surplus vegetables. She does not start with an app. She posts a short, honest question in two gardening groups asking how people handle a glut of courgettes, and she gets forty replies. Some say they give them away at the gate, some say they never bother, some say they would love a simple way to find neighbours.

She talks to eight of them. Five describe the same annoyance: they have extras at the same time as their neighbours, so everyone has too much of the same thing. Two have tried a messaging group and abandoned it. She builds a one-page description of a simple swap board for one site and shares it with the eight. Four say they would try it. One asks if it could cover the whole town.

She runs a manual version for two weeks in a single allotment group with a shared spreadsheet and a weekly message. People use it. Then she builds the smallest app. At every step, the next step was chosen by evidence from the step before, and nothing was built until the problem had been shown to be real.

Keep your notes in one place

A single document of dated conversation notes, test results and decisions saves you from re-learning the same lessons. Tag each note with who said it and whether it was a stranger, a friend or a possible customer. When you later write your public page, much of its wording will come from this file, in the customers' own phrases, which is exactly what makes copy ring true.

The short version

Decide what you need to learn, talk to people about their lives, watch what they already do, try the job by hand, test interest with an honest page and, if you can, ask for a real commitment. Rank evidence by how costly it was for the person to give, guard against false positives and negatives and be ready to stop. The cheapest product is the one you did not need to build.

Questions and answers

What does validating an idea mean?
Finding evidence, before you build, that real people have the problem and would use or pay for your solution.
Is a landing page test honest?
It can be, if the page is clear that the product is not available yet and you invite people to join a list or pre-order with honest terms.
How many people do I need to talk to?
Enough to see a pattern. Many founders start with about ten, then keep going until answers repeat.
What counts as strong evidence?
Behaviour: people paying, booking time, giving you their data or using a manual version. Compliments are weak evidence.
What if the idea fails the test?
That is a good result. You saved months, and the conversations usually point to a better idea.

Sources

Related guides

Help centre