Demand test
Fake Door Test Guide — Measure Demand Before You Build
A fake door test (also called a smoke test) shows a promise and a button — waitlist, pre-order, or “get early access” — before the product is ready. The point is not to trick people. The point is to measure whether anyone will take a concrete next step toward the outcome you named.
Last updated: September 6, 2026
Direct answer
What is a fake door test?
A fake door test is a demand experiment: you present the value proposition and a single action, then observe how many qualified people click, sign up, or pay intent — without shipping the full product. It answers “will anyone reach for this?” not “can we engineer it?”
Key takeaways
What to remember
- One audience, one promise, one CTA — extra options destroy the signal.
- Traffic quality matters more than traffic volume. Wrong visitors produce false negatives.
- A click is weaker than a signup; a signup is weaker than a deposit or paid pre-order.
- Set a stop rule before you start. Otherwise every weak result becomes “we need more traffic.”
- After the test, talk to responders. Behaviour opens the door; conversation explains it.
When to use it
When a fake door helps — and when it does not
Use a fake door when the expensive unknown is demand or messaging, not technical feasibility. Skip it when you still cannot name who pays or what pain you are naming.
Use it when
You can name a specific ICP, a concrete outcome, and a reachable channel. You want a cheap read on interest before weeks of build work.
Skip it when
The risk is “can we deliver?” rather than “does anyone care?” In that case a concierge MVP or narrow prototype teaches more than a button.
How to run it
Six steps for a clean fake door
Keep the test boring and measurable. Clever pages hide weak demand.
- 1
Write one falsifiable bet
Example shape: “People who [role] and [recent pain] will leave an email (or pay X) for [named outcome] within [timebox].” If you cannot write the bet, you are not ready to test.
- 2
Freeze the ICP and the channel
Pick one segment and one place you can actually reach — a community, a search query, a short outreach list. Mixing channels mixes conclusions.
- 3
Build the thinnest honest page
Headline = who + pain + outcome. One short explanation. One CTA. Say what happens next (waitlist, reply, or manual follow-up). Do not invent features you will not deliver soon.
- 4
Drive a small, intentional sample
Prefer dozens of right people over hundreds of random visitors. Track source so you can discard junk traffic later.
- 5
Capture behaviour, then follow up
Log clicks, signups, and any money signal. Message responders within a day: what they hoped for, what they do today, what would make them pay.
- 6
Decide against the stop rule
Compare results to the bet you wrote on day one. Continue, rewrite the promise, change the segment, or stop — write the decision in one sentence.
How to read results
What counts as signal
Treat numbers as preliminary standards for your next move — not proof of a market.
Stronger signal
Qualified people take the CTA and answer a follow-up with a recent, concrete pain story — or leave a deposit.
Weak but useful signal
Clicks without signup, or polite signup with no reply. Useful for messaging, not for “build the product.”
Failure signal
Right-looking traffic ignores the CTA, or responders describe a different problem than the one on the page.
Invalid test
Most traffic is off-ICP, the CTA is ambiguous, or you changed the headline mid-test without restarting the count.
There is no universal conversion rate that “proves” an idea. Compare yourself to the stop rule you set for this audience and channel — not to someone else’s dashboard screenshot.
Checklist
Copy-ready run checklist
- Bet written in one sentence with a timebox
- ICP named in plain language (not “everyone”)
- Single channel chosen and recorded
- Page has one H1, one promise, one CTA
- Next step after CTA is honest and fast
- Analytics or spreadsheet ready before traffic starts
- Follow-up questions prepared for responders
- Stop / continue rule written before launch
Common mistakes
What usually ruins the test
Building a brochure instead of a bet
Five CTAs and a feature grid measure design preference, not demand.
Calling friends and calling it validation
Warm contacts inflate signup rates. Prefer strangers who match the ICP.
Moving the goalposts after a weak week
If results miss the stop rule, change the experiment — do not quietly lower the bar.
Never talking to people who clicked
Without follow-up, you only know that a headline was interesting — not that the problem is real.
Promising a product you cannot follow through on
Be clear that this is early access or a waitlist. Collect only what you need; do not harvest personal data you will not use.
Continue or stop
What to do after the timebox
Continue validating when
Qualified responders take the CTA and their follow-up stories match the page promise. Next: deeper interviews, a pricing ask, or a concierge delivery for a few of them.
Stop or reframe when
Even targeted traffic will not act, or every conversation points at a different pain. Do not “build anyway and market harder.”
Sources
Where these ideas come from
- The Lean Startup principles (Eric Ries) — Smoke tests and early demand experiments before full build.
- The Mom Test (Rob Fitzpatrick) — How to talk to responders without leading them into polite yeses.
With Startup MRI
Use the report to pick the right test
If distribution or validation-cost risk is weakest on your Startup MRI report, a fake door is often the cheapest next experiment. Run the test, then bring what happened back into Evidence → Re-evaluate on the same report.
Analyze my idea →FAQ
Questions founders actually ask
Is a fake door test unethical?
It is unethical if you imply the product is ready and disappear. It is honest if you state that this is early access or a waitlist and you follow up. Collect the minimum contact detail you need.
How is this different from a normal landing page?
A marketing landing page often assumes the product exists. A fake door is an experiment with a written bet, a timebox, and a stop rule — the page is the instrument, not the brand site.
Do I need paid ads?
No. Paid ads are one channel. Outreach, communities, and existing lists can work if the people match the ICP. What matters is knowing where traffic came from.
What if nobody clicks?
First check traffic quality and whether the promise is clear. If the right people still ignore it, treat that as evidence against the current bet — rewrite the promise or the segment before you build.
Should I ask for payment on the fake door?
A small deposit or pre-order is stronger than an email. Only do this if you can deliver something useful soon (even manually). Otherwise a waitlist plus fast interviews is safer.
How does this connect to Startup MRI?
The report highlights which risk is weakest. When that risk is “will anyone reach for this?”, the recommended experiment is often a landing or channel test — this guide is how to run that test without turning it into a full marketing site.
Know which risk to test first
Startup MRI scores the weak assumptions, then points you at a concrete experiment — including demand tests like this one.
Generate a Startup MRI report →