How do you validate demand before you build?

What does it mean to validate demand?

Validating demand means proving that people will do something, not just say something. You show them the offer, ask for a real step like an email, a click or a payment, and count how many take it.

This is different from checking that the problem is real. Talking to people tells you the pain exists. A demand test tells you they will act on it. If you have not done the interviews yet, start with our guide on how to validate a startup idea, then come back here.

Why bother? Many startups fail because nobody needed what they built. And what people say is a weak guide to what they do. Plenty of people will tell you they would use something, then never touch it once it ships.

Think of demand tests as a ladder. Each step asks people for more, so each step gives you a stronger answer. Start at the bottom, where tests are fast and cheap, and climb only when the step below works.

• Landing page and waitlist

  What people do:

  Leave an email

  What it proves:

  Enough interest to raise a hand

  Number to watch:

  Visitors who sign up, split by where they came from

• Waitlist with a price shown

  What people do:

  See a price, pick a plan, then leave an email

  What it proves:

  They saw what it will be and kept going

  Number to watch:

  Share of visitors who click a paid plan

• Fake door test

  What people do:

  Click a button for something not built yet

  What it proves:

  They want this exact thing

  Number to watch:

  Click rate on the button

• Pre-order or pre-sale

  What people do:

  Pay, or put down a deposit, before it exists

  What it proves:

  They will pay for it

  Number to watch:

  Share of people asked who pay

• Paid pilot

  What people do:

  Sign up and pay for a short trial on real work

  What it proves:

  A business will pay to try it

  Number to watch:

  Pilots that turn into ongoing deals

Is a waitlist enough to validate demand?

No. A waitlist on its own shows interest, not demand. A signup takes a few seconds and promises nothing, so a long list can still be full of people who will never pay.

The count is the start of validation, not the end of it. A huge list of people who thought the product was free can be worth less than a small list of people with a painful problem they already pay to solve.

Waitlists also go stale. A waitlist guide from a16z speedrun, citing Lenny Rachitsky's 2022 data, found that 5% to 25% of waitlisted people became paying customers. It averaged around 20% when people got off the list in under a month, and fell below 10% when they waited more than three months.

When Nomiki Petrolla, Theanna's founder, built Theanna's waitlist, the page said plainly what it was, who it was for and why she was building it. Each signup got a thank-you email, then an update every week, so the list stayed warm until launch.

To make a waitlist mean something:

• Ask one or two questions when people join, like what they use today and how big the problem is.
• Show a price before the email form, so casual visitors drop off.
• Email the list fast and ask for the next step, like a call, a pre-order or a pilot.
• Talk to a handful of signups to learn why they joined.
• Let people in within weeks, not months.

How many signups do you need to validate demand?

There is no magic number. What matters is who signed up and what share of them take the next step, not the size of the list.

Still, you need enough people to trust the result. Send enough visitors to a test page that a few clicks either way cannot flip your answer. If the count is still small, keep the test running.

Signup rates swing a lot based on where visitors come from. Friends and followers sign up far more often than strangers from an ad. That is why published averages make poor targets. Track your own rate, split by source, and compare it to itself over time.

A simple way to set your own number is to work backward from the money. If you want 20 paying customers and roughly 1 in 5 of a fresh list pays, you need about 100 signups who fit your customer and who you let in quickly.

What is a fake door test, and how do you run one?

A fake door test is a button, menu item, pricing tier or ad for something you have not built yet. You count how many people click it. When they do, you tell them the truth: it is coming soon, and they can join early access.

A smoke test is the same idea for a whole product. You build one page that describes the product as if it exists, add a signup or pre-order button, and send people to it. A fake door usually tests one feature inside a product you already have.

Theanna started this way. The first version was a list in a Notion document. Users kept asking for more, so the team shipped a simple fake agent before building the real one. People used it so much that they broke it. That usage, not anyone's opinion, is what told the team to build the real agent.

Run it for a set window, and pick the end date before you start. Run it longer if you have little traffic, so a few clicks either way do not decide it for you.

Set your pass mark before you start. There is no universal number, so compare the click rate to a button or feature people already use. A few rules for running it:

• If the click rate is close to a feature people already use, move to the next step on the ladder.
• If the result is unclear, run it longer or talk to the people who clicked.
• Always be honest after the click. Never leave people feeling tricked, and do not show the same fake door to the same person again and again.

How do you use pre-sales to validate demand?

Ask people to pay before you build. Money is the strongest demand signal there is. One pre-order tells you more than a pile of likes.

You do not need a finished product to ask. Buffer started with a two-page test: a page describing the app and an email form. Then the founder added a pricing page in between. People kept clicking through, and a small number clicked paid plans. That was enough to start building, and Buffer had its first paying customer within four days of launch.

Theanna's founder did the same with a service. Before there was any software, Nomiki sold coaching for about $7,000 for six months. What she taught and what founders asked for in those sessions became the plan for the product.

Ways to ask for money early:

• Pre-order: take full payment now, with a clear ship date and a full refund if you miss it.
• Deposit: take a small payment to hold a spot, taken off the full price later.
• Founding customer deal: a lower price for the first group, in return for feedback and patience.
• Paid pilot: for businesses, a short paid trial with a clear goal and a date to decide on a full deal.
• Be plain about what exists today, when the rest arrives and how refunds work.

How do I run a demand test as a short experiment?

1. 01

   Write down what you are testing

   One sentence: who you think wants this, and what you will ask them to do. For example, "Shop owners on Shopify will pre-order a returns tool for $29 a month."

2. 02

   Pick the action and your pass mark first

   Choose one action to count, like signups, clicks or payments. Write down the number that means yes before any traffic arrives, so you cannot move the goal later.

3. 03

   Set a length and stick to it

   Pick an end date before any traffic arrives, and give it longer if your traffic is low. Do not stop early because the first day looks great or terrible.

4. 04

   Build the smallest thing that asks

   One page, one button or one checkout link. No product, no logo work, no long setup. The test only needs to ask the question clearly.

5. 05

   Send the right people to it

   Reach the people you wrote down in step one, in the places they already spend time. Tag each source so you can tell which visitors acted, and send enough people to trust the count. Testing more than one channel? Run each as its own test. For Theanna's waitlist, Nomiki ran five at once, one each for LinkedIn, TikTok, Instagram, her email list and a lead list, so each result was clear.

6. 06

   Count, then talk to the people who acted

   Check your number against your pass mark. Then reach out to a few people who signed up, clicked or paid, and ask why. The reasons tell you what to say next.

7. 07

   Decide, then climb or change

   If it passed, move up the ladder and ask for more, like a pre-order instead of an email. If it failed, change one thing, such as the audience, the offer or the price, and run it again. Most tests will not work, and that is normal.

What are false signals of demand?

A false signal is anything that feels like demand but asks nothing of the person. It is easy to give and easy to forget.

Big numbers fool everyone. Nomiki once had a video shown 300,000 times, and it turned into nothing. In a Meta ads test for a Theanna webinar, making the form easier pushed signups from 148 to 653, but fewer than 1 in 10 people showed up. The easier the step, the less it tells you.

Watch out for these:

• Likes, shares and comments on a post about your idea.
• Compliments like "this is so cool" or "I would totally use that."
• Free signups who never open your emails or come back.
• Friends and family who sign up to be nice.
• A big list from a giveaway, where people joined for the prize.
• A traffic spike from a launch site that does not match your customer.
• Survey answers about what people might do someday.
• The real signals are the ones that ask for something: time on a call, a click past a price, a card number, a signed pilot. When in doubt, ask for the next step and see who takes it.