How to do it
Build one page that makes a specific promise at a specific price
Vagueness ruins the test. 'Better analytics' measures nothing; 'weekly churn alerts in Slack for ₹1,500/month' measures something. The price must be on the page — a click on a free thing tells you nothing about a paid one.
Send traffic that resembles your real buyer
The conversion rate is meaningless without knowing who saw it. Three hundred people from a community of practitioners and three hundred from a general startup newsletter will produce different numbers for the same page, and only one of them is your market.
Measure the checkout click, not the signup
The meaningful event is reaching the payment step, because that is the last action before parting with money. An email capture measures curiosity. Track both if you like, but decide on the checkout number.
Tell the truth on the next screen
Take nobody's money for something that does not exist. The screen after the button says plainly that you are testing demand, when it will be ready, and offers a real way to be told. Founders who skip this get a clean number and a reputation problem in the exact community they need.
Worked example
The honesty problem, and how to handle it
The test only works because people believe they can buy. That belief is also what makes it possible to mislead them, so the design of the screen after the button is not an afterthought — it is the part that decides whether the test was ethical.
What works: a plain statement that it is not built yet, a real date or a range, and one option to be told when it is. What does not work: taking a card number, showing a fake error, or a page that implies an outage rather than an absence. Founders who take the shortcut get a clean number and lose the first twenty people who would have been their earliest customers.
Reading the number honestly
A conversion rate means nothing without the traffic source attached to it. Two percent from a tight community is a strong signal; two percent from broad paid traffic is close to noise. Always report the pair.
Under two hundred targeted visitors, do not draw a conclusion. At that volume a single extra click moves the rate by half a percentage point, and founders routinely make six-month decisions on the difference between four clicks and seven.
What goes wrong
- Hiding the price, which turns the test into a curiosity survey
- Running it on untargeted traffic and then reading the conversion rate as if it meant something
- Stopping at under 200 targeted visitors, where a handful of clicks is noise
- Taking payment for something that does not exist
- Treating the email captures as the result — they are the consolation prize
Is your number plausible?
Each check catches a different way the arithmetic can be right and the answer still wrong.
- Under 200 targeted visitors, do not draw a conclusion — single clicks move the rate by half a percent
- Compare against the same traffic source, not an industry average; cold ads and warm community traffic are different tests
- If the checkout rate is above about 10%, check whether the page promised something you cannot actually deliver
Related answers
Is a fake door test dishonest?
Not if the next screen tells the truth immediately. The dishonest version takes money or hides the fact that the product does not exist. The honest version measures intent and then explains itself, which most people respond to well — several will ask to be told when it ships.
Should I charge on the fake door?
Do not take payment for something that does not exist. If you want a stronger signal than a checkout click, offer a genuine pre-order with a clear delivery date and an unconditional refund — that is a real transaction, not a test.
What conversion rate means it worked?
Compare against the same traffic source, not a benchmark. The useful question is whether the rate is high enough to justify the next, more expensive test — not whether it clears some published average.