Ways to run the test
What is a fake door test and how do I run one?
Updated 19 August 2026 · Thresholds read live from the verdict engine
Short answer
A fake door test puts a button in front of people for something you have not built, then counts how many press it and what happens when they do. Done properly the visitor learns straight away that the thing is coming, and nobody is charged for a product that does not exist. The press rate is your demand signal, and with a price attached it is the closest pre-build proxy for revenue there is.
What a fake door test is
The name comes from painting a door on a wall to see how many people try the handle. Online it is a page, a button or a menu item that promises something you have not built, wired to a screen that tells the truth once the person has acted.
The technique is old and well documented. Zappos began as a founder photographing shoes in local shops and buying them at retail whenever an order came in, which tested whether people would buy footwear online before any inventory existed. Dropbox showed a three-minute video of a product that was barely working and grew its beta list overnight. Buffer put up a two-page site where the second page listed prices that could not be paid, and counted how many people clicked through to a plan.
All three measured the same thing: what people do when the offer is in front of them, rather than what they say when asked.
The three kinds, and which one you need
1. The pre-launch door
A landing page for a product that does not exist, ending in an email capture or a checkout. This is the one most founders mean. It tests the promise and the audience at the same time, which is a strength when it passes and a nuisance when it fails, because a failure could be either.
2. The in-product door
A feature that is not built yet, exposed as a menu item or a button inside a product that already has users. Pressing it opens a panel explaining that the feature is coming and offering to email them. Cheap, precise, and the fastest way to rank a roadmap. It only works if you already have traffic.
3. The price door
The offer is real and so is the pricing page, but the checkout stops short of a charge. Either it stores a card without taking money, or it takes a refundable pre-order. This measures the thing you care about most, which is commitment at a number.
| Kind | What it measures | Traffic needed | Best for |
|---|---|---|---|
| Pre-launch door | Promise plus audience | 300 to 500 cold visitors | A new product idea |
| In-product door | Feature demand among existing users | A few hundred sessions | Roadmap decisions |
| Price door | Willingness to commit at a price | 300 to 500 cold visitors | Deciding whether to build and at what price |
Running one in an afternoon
- Write the offer as one outcome. Name what the person has afterwards. Everything else on the page supports that sentence.
- Put the price in view. The number you plan to charge, in the shape you plan to charge it. A door with no price measures curiosity.
- Build the door. A single button, above the fold, saying what happens next in the visitor's words ("Get early access", "Start the trial").
- Write the room behind the door. The screen after the click does the honest work: it says the product is launching soon, offers to email them, and thanks them properly. Never leave a dead end or a 404.
- Track the click and the completion separately. Half of what you learn is in the gap between them.
- Send cold traffic. $40 a day for five days, three creatives, one country. Warm traffic makes every door look popular.
Where the ethics line sits
The test is honest when nobody ends up worse off for having pressed the button. That gives you a workable rule with three parts.
- Tell them immediately. The disclosure belongs on the screen right after the click, not in a footer or a confirmation email sent later.
- Do not take money you might keep for nothing. Storing a card without charging is fine when the screen says so. Charging for a launch date you cannot commit to is not. If you take pre-orders, refund every one of them the day you decide against building.
- Deliver the consolation. The email you promised, the founding price you offered, the update when you decide. People forgive a door that turns out to be paint. They do not forgive silence.
There is a commercial reason to hold that line as well. Ad platforms enforce policies about promoting products that people cannot obtain, and repeated rejections put the ad account itself at risk. A page that says the launch is upcoming reads as a pre-order to a reviewer. A page implying the app is on the store today does not.
What a fake door cannot tell you
- Whether they would still want it in a month. Intent decays, and a door measures a single moment of attention.
- Whether the product can be built to match the promise. The better your copy, the wider the gap you have to close later.
- What people will pay at scale. The first audience you buy is the most responsive one. Costs rise and rates fall as you spend more.
- Why they left. The number tells you that they did. Session recordings, a one-question exit survey, or ten quick messages to the people who gave an email tell you why.
Worked example
A door that failed usefully
A team tested a $59 a year plant-care app. The first angle, "never kill another houseplant", pulled 480 visitors and 4 checkouts, which is 0.83%. Below the kill line.
Rather than dropping the idea they changed the promise to "know exactly when to water, from a photo" and reran it for another $200. Same product, same price, 2.9% conversion on 410 visitors. The first test did not say the idea was dead. It said that sentence was.
The numbers
A fake door with a price gets judged on one number: completed checkouts divided by visitors. These bands come from the verdict engine.
- Funnel starts before any verdict
- 30
- Below this, don't build
- 1.0%
- At or above this, build
- 2.0%
- Visitors before a segment counts
- 50
Under 1% means optimisation will not bridge the gap at this price and angle. Between 1% and 2% the offer is inside optimisation reach, so change one thing and rerun. At 2% you match shipped web2app funnels before any tuning. Nothing is called under 30 funnel starts.
| Result | Rate | 95% range | Confidence | Reads as |
|---|---|---|---|---|
| 4 / 480The failed plant-care angle | 0.83% | 0.32% to 2.12% | medium | Don't build |
| 12 / 410The rewritten angle | 2.93% | 1.68% to 5.05% | medium | Build |
| 6 / 300 | 2.00% | 0.92% to 4.29% | medium | Build |
| 15 / 1,000 | 1.50% | 0.91% to 2.46% | medium | Promising |
How VerifyToLaunch does this
The funnel VerifyToLaunch generates is a price door with the disclosure built in. Six screens carry your promise and your price, the checkout asks them to pick a plan and confirm at it with no card involved, and the screen after it tells the visitor plainly that the product is in pre-launch and what they get for having signed up. Emails collected on the way are yours to export.
You still write the promise, since nobody else knows your market, and you still buy the traffic in your own ad account. If the honest disclosure is the part you were hoping to skip, this is the wrong tool.
Common questions
- Is a fake door test legal?
- Running one is legal in the usual jurisdictions as long as you do not take money under false pretences and you handle collected emails under your local privacy law. The risks that bite in practice are advertising policy and reputation, both of which are managed by disclosing on the screen right after the click.
- Should I collect card details or just emails?
- Card details, if you can. An email costs a visitor nothing and measures curiosity; a card costs trust and measures commitment. Collect it through a Stripe setup intent so nothing is charged, and say that on the screen.
- How many visitors does a fake door test need?
- Between 300 and 500 cold visitors settles a result that is clearly above or clearly below the bar. A rate near 1% to 2% can still be undecided at 1,000 visitors, which is a reason to change the offer rather than buy more traffic.
- What do I do with the emails if I do not build it?
- Send one honest message saying you decided not to build, and why. Offer them the alternative you would recommend. People remember that, and some of them become the first customers of whatever you do build.
Read next
- How do I run a smoke test with a landing page?The page block by block, four events to track, and how to read every drop-off.
- How do I validate an app idea?The whole method: pick a commitment that costs the visitor something, buy cold traffic, count who commits.
- How do I find out if people will actually pay?Three ways to collect a paying signal before you build, and the CAC maths behind the rate.
- What conversion rate should I expect?Benchmarks by traffic type and ask, plus how to diagnose a bad rate stage by stage.
Run the door without building it
The generated funnel is a price door with the disclosure already written: your promise, your price, a checkout that never charges, and a verdict computed against the thresholds on this page.
One-time licence, 30-day money-back guarantee