Fake Doors That Don't Lie: How to Run Landing Page Tests Without Fooling Yourself

Fake doors work — until founders misuse them

Fake door tests are one of the fastest ways to test demand without building.

They're also one of the most abused.

Most founders run fake doors like this:

1. Create a landing page
2. Drive traffic
3. Count signups
4. Declare validation

That's not a test.

That's a vibe check.

A real fake door test is not about how many people sign up — it's about what people are willing to do next.

What a fake door is actually for

A fake door answers one question:

"If this existed today, what behavior would it trigger?"

Not:

• "Do people like the idea?"
• "Is the copy compelling?"
• "Can I get cheap clicks?"

But:

• "Will someone take a meaningful step toward solving this problem?"

The door is fake. The decision must be real.

The three-layer fake door model

At Forge, we design fake doors with three layers.

Most founders stop at layer one.

Layer 1: Message resonance

This is what most people test.

Metrics:

• Click-through rate
• Scroll depth
• Basic signups

Use this to compare messaging, not demand.

If nobody clicks, the problem may not be real — or you're talking to the wrong audience.

Layer 2: Intent escalation

This is where signal starts.

Add an action that requires effort:

• "Book a demo"
• "Apply for early access"
• "Tell us about your use case"

Now you're testing:

"Is this worth my time?"

Expect drop-off. Drop-off is not failure — it's clarity.

Layer 3: Commitment stress test

This is where truth lives.

Ask for:
• money
• scheduling
• real data
• public commitment

Examples:
• Pre-order at a discounted price
• Join a paid pilot
• Upload sample data
• Reserve a slot in a limited cohort

Even one yes here beats 1,000 emails.

How AI helps (and how it hurts)

AI makes fake doors easier than ever:

• Copy in seconds
• Designs in minutes
• Variations at scale

That's the upside.

The downside is founders optimize for persuasion, not truth.

AI-written copy is very good at making things sound valuable — even when they aren't.

Your job is not to convince users. Your job is to let them convince you.

Designing fake doors that reveal reality

Here are four rules we use at Forge.

1. Price early (even if it's wrong)

Free fake doors lie.

A price, any price, forces a real decision.

You're not testing price accuracy yet. You're testing willingness to pay at all.

2. Be specific, not ambitious

"AI-powered platform to transform your workflow" tells you nothing.

"Automate weekly compliance reports for 10–50 person fintech teams" gives users something concrete to react to.

Specificity creates signal.

3. Make the next step slightly uncomfortable

Good fake doors create friction on purpose.

If the next step feels trivial, the signal is trivial.

4. Follow up manually

Automation hides nuance.

Talk to the people who cross each threshold:

• Why did they click?
• What almost stopped them?
• What would make this a must-have?

These conversations are worth more than traffic.

A common mistake: testing too many things at once

Founders often overload fake doors:

• multiple audiences
• multiple problems
• multiple solutions

Then they can't tell what worked.

One fake door = one hypothesis.

If it fails, that's progress.

When a fake door has "passed"

A fake door hasn't passed when:

• it hits a conversion benchmark
• it "feels promising"
• your friends like it

It has passed when:

• at least a few people take a Tier 3 commitment action
• you understand why they did
• you can articulate who this is for and who it isn't

At that point, you've earned the right to build something.