Publish a one-page site that describes your proposed app, clearly says it is in development, and offers one action such as "join the list" or "reserve with a deposit". Send at least 100 to 300 people who fit your customer, count how many act, and compare the rate with a stop rule you wrote before launch.

This is for founders who want a demand signal before they build. The running example is TeamPot, an imagined app for splitting youth football team costs between parents.

In short:

  • A smoke test measures interest in a promise, not repeat use.
  • One button per page. Every extra link splits attention.
  • At 100 visitors, a 5% rate could really be 2% to 11%.
  • In the UK and EU, analytics cookies usually need consent first.

A smoke test answers some questions, not all

A smoke test tells you whether your promise attracts the people you think it does. It can't tell you whether they'll keep using the product.

Use a page to test interest and price, and a small working app to test usage.

Question Smoke test answers it?
Do my target customers care? Yes, roughly
Which of two promises works better? Yes, with enough traffic
Will people pay this price? Partly, with a deposit button
Will people use it every week? No

To pick which assumption to test first, see how to validate an app idea in a weekend.

The page needs five parts and one button

A smoke test page is short. Build it in this order:

  1. Write a headline that names the outcome for one person: "Split team costs without chasing parents."
  2. Add one line that says how: "Each parent sees what they owe and pays from a link."
  3. List three benefits in the customer's own words, from interviews or reviews.
  4. Show the product with a screen design or a short screen recording.
  5. Place one action, with a line under it saying what happens next.

"We'll email you when the first version is ready, at most once a month" sets expectations. Honesty here builds trust and protects you.

Choose the action that matches your risk

The more an action costs the visitor, the stronger the signal it gives.

Action Visitor effort What it tells you
Email sign-up Low Interest
Two-question form Medium Interest and who they are
Book a 15-minute call High Strong interest, plus an interview
Refundable deposit Highest Willingness to pay

For a deposit, Stripe Payment Links give you a hosted payment page from a link. Say plainly that the product is in development and that you refund on request.

Count events, not page views

Install analytics before you share the link. Track three things: views by traffic source, clicks on the main button, and completed actions.

Tag every link you share with a source, such as ?utm_source=reddit. Google Analytics 4 handles this for free. So does PostHog. If you run Meta ads, the Meta Pixel reports sign-ups back to the campaign.

In the UK and EU, analytics and ad pixels that set cookies generally need consent. The UK regulator's guide to cookies and similar technologies explains the rules. Expect some visitors to decline, which lowers what you can count.

Send visitors who fit the customer

A hundred of the right visitors tell you more than 500 random ones. Four sources work for a first test:

  • Search ads reach people looking for a solution now.
  • Social ads reach people who match a profile but aren't looking.
  • Communities bring high-intent visitors if you follow their rules.
  • Direct messages to interviewees and their friends.

Report each source separately. A 12% rate from a parents' group and 1% from cold ads are two different facts. Averaging them hides both.

100 visitors catch failures, not small wins

Small samples swing a lot, so you need about 300 visitors before a sign-up rate means much. Here is the plausible range around a measured 5% rate, using a 95% confidence interval (Wilson method).

Visitors Sign-ups Plausible true rate
100 5 About 2% to 11%
200 10 About 3% to 9%
300 15 About 3% to 8%
1,000 50 About 4% to 6.5%

So 100 visitors is enough to see zero sign-ups or a clear hit. Comparing two headlines needs far more. Evan Miller's sample size calculator shows how many per version.

Read the result against rules set first

Write three bands before launch. For example: stop under 3% after 300 qualified visits, rerun with one change between 3% and 8%, and continue over 8% with a few replies or deposits.

These numbers are illustrations. A deposit at 2% can beat an email at 10%. Then read every reply, and email sign-ups one question: "What made you sign up?"

Five mistakes make smoke tests lie: friends in the numbers, a slow page (check it with PageSpeed Insights), a vague promise, too many links, and moving the bar after seeing the data.

Build the test page in GrowApps

In GrowApps you describe the page in a chat, approve the screens, and publish it at a growapps.cloud address. A findability checklist runs at every publish, and sites meant to be found ship real HTML next to the Flutter app.

You connect your own PostHog or Google Analytics account and a Stripe payment link in a browser window GrowApps opens for you (sign-in, data, AI and payments). If the test says continue, the page becomes the front door of the app in the same project.

Write your headline and stop rule now

Write one headline that names the outcome for your customer, and one stop rule with a visitor count and a percentage. If you can't write the headline in under ten words, talk to two more customers first. Then plan what to build next with our MVP app development guide.

Frequently asked questions

Is a smoke test ethical?

Yes, if you're honest. Say the product is in development, take money only with a clear refund promise, and send sign-ups a real update. A page that pretends a product exists and then goes silent damages trust.

What is the difference between a smoke test and a fake door test?

They're close. A smoke test is usually a whole landing page for a product that doesn't exist yet. A fake door test is usually a button for a feature inside an existing product, counting clicks before the feature is built.

Sources and further reading