How to run a fake door test without misleading anyone
A fake door test, sometimes called a painted door test, puts a button for something you haven't built in front of real visitors and counts who presses it. Done carelessly it feels like a trick. Done honestly it's a polite question: "would you want this?", answered by behaviour rather than opinion. Here's how to run one that respects the people taking part.
Where the line is
The difference between a test and a deception is what happens after the click. An honest test tells the visitor straight away that the thing isn't available yet, explains why they're seeing the button, and offers a way to hear when it is. A deceptive one lets them fill in card details, or keeps them guessing, or pretends there's a technical fault.
- Never take money for something you can't deliver. If you want to test willingness to pay, run a real pre-order with real delivery dates instead.
- Never collect more than you need. An email address for a waitlist is enough.
- Keep it short. A door that stays painted for months annoys loyal visitors. Set an end date.
- Follow through. Write to everyone who joined, whether you build it or not.
Plan the test
Write down three things before you build anything: what you're testing ("a weekly meal-kit option for existing customers"), where the door appears (the pricing section, a navigation link, a product page), and what result would change your plans. A fake door gives a click rate on the door and a sign-up rate behind it. Decide the numbers that mean build, drop or rethink, and the number of visitors you'll wait for before reading them.
Building the door and the room behind it
- Build the room as its own small site. Anticode counts views per site rather than per page, so publishing the page behind the door as a separate project, on its own anticode.dev address, gives it a view count of its own.
- Add the door to your main site. Ask for it in a sentence: "add a Meal kits link to the navigation and a card for it in the services section, both linking to this address". The change request updates the build.
- Write the room plainly. Its headline says what the idea is and that it isn't available yet: "Weekly meal kits: we're deciding whether to make them." Then two or three lines on what it would be, and a newsletter sign-up for people who want to hear if it happens.
- Send sign-ups somewhere useful. Each one is saved in the dashboard, and a connector can pass it to Kit, Mailchimp or a Google Sheet.
- Check the wording on a phone, where most people will read it, and make sure nothing on the page suggests the product exists.
- Publish and note the date.
Don't point a buy button at a real checkout for this. Anticode's buy buttons open a payment link from your own provider, and a real link would take real money.
Measuring the result
Anticode's analytics count views per site per day, by source, without cookies. Views of the small site behind the door tell you how many people opened it; set them against your main site's views over the same days for a rough click rate. Sign-ups in the dashboard, against the door site's views, give the second rate. There are no click maps or per-button counts, and anyone who reaches the door site some other way (a shared link, a search) blurs the figure, so don't promote its address anywhere else during the test.
| What you see | What it probably means |
|---|---|
| Many opens, few sign-ups | The idea sounds interesting, but the page behind it didn't convince |
| Few opens, high sign-up rate | A small group cares a lot; the door may be badly placed |
| Few opens, few sign-ups | Not wanted in this form, by this audience |
| Many opens, many sign-ups | Worth a deeper test: interviews, then a real pre-order |
Closing the test
On the end date, remove the door with another change request, or replace the page with an update on what you decided. Email everyone who signed up with the outcome from your email tool; Anticode doesn't email your list. If you go ahead, they're your first customers. If you don't, a short honest note keeps their trust.
Limits
Anticode gives you the extra page, the sign-up form and views by source. It has no A/B testing, so you can't show the door to half your visitors, and no event tracking for individual buttons. For a test of a whole product rather than one feature, see validating an idea with a landing page; for the payment step that should follow a good result, taking pre-orders.
Questions people ask
Is a fake door test ethical?
It can be, if the page behind the door says plainly that the product isn't available, takes no money, asks only for an email, and you follow up with everyone who joined.
How long should a fake door test run?
Until you reach the visitor count you set in advance, and rarely more than a few weeks, so regular visitors don't keep meeting a door that leads nowhere.
What's the difference between a fake door test and a painted door test?
Nothing in practice; they're two names for the same method.