---
name: Fake door
slug: fake-door
category: pattern
status: published
created: 2026-08-28T00:00:00.000Z
modified: 2026-08-28T00:00:00.000Z
definition: A control that looks shipped and leads nowhere, put in the interface
  on purpose to count how many people want the feature behind it.
aliases:
  - name: fake door test
    source: community
  - name: fake door testing
    source: amplitude
  - name: painted door
    source: community
  - name: painted door test
    source: optimizely
tags: []
relations:
  contrastWith:
    - feature-gate
    - dark-pattern
  variantOf: []
  partOf: []
  seeAlso: []
implementations: []
sources:
  - title: "Optimizely optimization glossary: Painted door test"
    url: https://www.optimizely.com/optimization-glossary/painted-door-test/
  - title: "Amplitude: What is fake door testing?"
    url: https://amplitude.com/explore/experiment/fake-door-testing
demo: inline
exhibit: false
useWhen: measuring demand for a feature by shipping only its entrance
---

A fake door is an entrance with no room behind it. The button sits where the real one
would sit, worded and weighted like everything around it, and the only thing wired to it
is a counter. The reasoning is arithmetic: a door costs an afternoon and the room behind
it costs a quarter, so the cheap half is built first and the expensive half is built only
if enough people push on the door. Product teams borrow the vocabulary from
experimentation, where the same thing is called a painted door test.

Almost all of that is a decision about what to spend, and none of it is design. The one
part a designer composes is what happens on the press, and it is the whole of the term's
quality. A door that opens onto nothing at all, or onto a generic failure, has told the
reader the product is broken and taught them not to try again, which corrupts the very
number the test was run to collect. An honest reveal does three things in about two
lines: it says the feature does not exist yet, it says why the button is there, and it
offers something real to do about it, usually being told when it ships. Then it gets out
of the way and leaves the interface exactly as it was.

The line this word must not cross is the difference between an experiment and a lie. A
signposted door tells the truth the moment it is opened and asks for nothing the reader
would not have given. A door that takes a card number, an address, or an afternoon of
setup for a feature nobody has built has stopped measuring demand and started
extracting value under a false claim, which is why [dark pattern](/dark-pattern) is the
contrast here and never the parent. Two smaller costs are worth counting before shipping
one. A press measures interest in the label as much as demand for the thing, so a door
labelled with an aspiration reports on the wording. And a door left standing for months
becomes a permanent piece of the interface that lies every time it is used, so the test
needs an end date decided before it goes up.

The neighbouring word is [feature gate](/feature-gate), and the difference is sharp:
behind a gate the feature exists and works, and the reveal names the plan that unlocks
it, so the reader is being sold something real. Behind a fake door there is nothing to
sell yet, and the reveal has to admit it. A gate at least carries a lock and the name of
a plan, so the reader knows before pressing what they are up against. A fake door carries
nothing, which is why the reveal is doing all of the honest work this pattern has.