How to Run a Fake Door Test Without Annoying Customers

A fake door test is one of the cheapest ways to learn whether people want something before you build it. You show customers a door to a new feature. When they try to open it, you record the click and explain that it is coming. The click shows interest through behavior.

The risk is trust. Done carelessly, a fake door feels like a bait and switch. Done well, most customers barely mind. Here is how we run one.

What a fake door test is

A fake door test (also called a painted door test) adds an entry point for something that isn't built yet. That entry point can be:

• A button or menu item inside an existing product, such as "Export to PDF."

• A landing page for a new product, with a sign-up or "Get started" button.

• A pricing page that lists plans for a service you haven't launched.

Each version measures how many of the right people step toward the offer when they see it.

When to use a fake door test, and when to use something else

A fake door test answers one question well: will enough people reach for this? It can't tell you why they clicked. Pick the method that fits the question.

We usually run interviews first to learn the problem and customer language. A fake door then checks whether that language pulls real behavior. Then 5 to 8 interviews with people who clicked explain what they hoped to find. Our guide to how many user interviews you need covers those sample sizes. The full sequence sits inside our guide to validating a product idea.

A well-known example: Buffer's landing page test

Joel Gascoigne, the founder of Buffer, described an early test in 2011. His first page explained the idea and let interested visitors leave an email. He then added a pricing page in between. In his words, that step tested pricing by showing which plan people clicked, and tested demand further. People kept clicking through, so he built the product. Each step asked a little more of the visitor and gave a sharper signal.

How to set up a fake door test in six steps

1. Write the hypothesis. Name the customer, the offer, and the behavior you expect.

2. Set the success threshold before launch, and share it.

3. Choose the door. Put it where the right people already are.

4. Write the label in the words customers used in interviews.

5. Build the post-click message (see the next section) and a simple waitlist form.

6. Track three numbers: views, clicks, and waitlist sign-ups.

What to show after the click

Amplitude's guidance is to disclose the test right after the click, explain it, give a rough timeline, offer a way to register interest, and make it easy to leave.

A good message is short and honest:

Rent reminders are on the way. We're deciding what to build next, and your click tells us this matters. Want early access? Join the list and we'll email you when it's ready. [Join the waitlist] [Back to my dashboard]

Three things make this work:

• It admits the feature isn't ready. No error page, no fake loading screen.

• It gives a reason. People accept "we're deciding what to build" far more easily than silence.

• It offers something. Early access turns a dead end into an invitation.

How to set a success threshold for a fake door test

Without a threshold set in advance, any result can be explained away. Here is a hypothetical example. A software company believes a new feature is worth building if at least 4% of active users who see the button click it, and at least 1 in 4 of those clickers join the waitlist. They also set a stop line: under 2% click, the idea goes back to discovery. Anything in between means a second test. Base thresholds on similar features, the customers you need to justify the build, and the cost of being wrong.

How much traffic you need

A click rate from 40 visitors is noise. The margin of error for a rate at 95% confidence is 1.96 × √(p × (1 − p) ÷ n), where p is the click rate and n is the number of people who saw the door.

If the true click rate is around 5%:

• 200 viewers: about ±3 points, so 5% could really be 2% to 8%.

• 500 viewers: about ±2 points, so 3% to 7%.

• 1,000 viewers: about ±1.4 points, so roughly 3.6% to 6.4%.

For example, at 500 viewers: 1.96 × √(0.05 × 0.95 ÷ 500) = 1.96 × 0.0097 = 0.019, or about 2 points.

For most go or stop decisions, a few hundred viewers per version is enough. If your threshold sits close to the expected rate, plan for 1,000 or more per version.

Ethics and trust

A fake door is a small promise you can't keep yet. Keep its cost low:

• Limit exposure to a slice of relevant customers, for a set period.

• Run one or two at a time. More start to feel like a pattern.

• Never take money. If you test a pricing page, stop at plan selection or the waitlist.

• Follow up. Tell the waitlist what you decided.

• Take extra care in regulated settings such as healthcare or finance, and check with legal.

Common fake door test mistakes

• Testing the wrong audience. Clicks from people who would never buy look good and mean little.

• Vague labels. If people don't understand the offer, a low click rate tells you nothing.

• Moving the goalposts. Changing the threshold after results arrive turns a test into a sales pitch.

• Treating clicks as revenue. Interest is a first signal. Willingness to pay needs its own test, such as the ones in our guide to pricing due diligence.

• Skipping follow-up interviews. The click shows what happened. Conversations show why.

Frequently asked questions

Is a fake door test the same as a painted door test?

Yes. The two names describe the same method: an entry point for something not yet built, used to measure interest.

Will a fake door test annoy customers?

It can if the message after the click is vague or hidden. An honest note, a reason, and an offer such as early access keep most customers on side.

How long should a fake door test run?

Run it until enough people have seen the door to give a stable rate, often a few hundred per version. For many products that takes one to three weeks.

What is a good click rate for a fake door test?

There is no universal number. Set your own threshold before launch based on similar features, the number of customers you need, and the cost of building.

Should we follow up with people who clicked?

Yes. Interview 5 to 8 of them to learn what they expected and whether the real product would meet that need.

Plan your next fake door test with us

Our Innovation Retainer gives your team a standing research partner. We pick the right test, set the threshold, write the post-click message, run the follow-up interviews, and deliver a clear build or stop recommendation. See how it works or start a conversation.

Sources

• Joel Gascoigne, "Idea to Paying Customers in 7 Weeks: How We Did It," Buffer, 2011

• Amplitude, "What Is Fake Door Testing: Methods and Best Practices," Amplitude, accessed 2026

• Margin of error figures calculated with the standard formula for a proportion at 95% confidence (1.96 × √(p × (1 − p) ÷ n))