Smoke Tests Explained: Measure Demand Before You Build

A smoke test measures demand for a product that doesn't exist yet, using a fake landing page, a disabled button, a waitlist, or a short video instead of working software. You show people the pitch and count who takes a real next step — a click that goes somewhere, an email address, a deposit — not who says something encouraging about the idea.

Quick Answer: Run a smoke test by putting a real offer — a landing page, a waitlist form, a "get notified" button that doesn't charge anyone yet — in front of real traffic, then measure the fraction who take a costly next step instead of the number who say nice things. Four patterns cover almost every situation: fake door, landing page plus ads, waitlist, and explainer video. Each trades cost, speed, and signal strength differently, so which one fits depends on whether you already have a product or an audience to test against.

Prerequisites: What You Need Before You Run One

A smoke test is one experiment pulled from the broader lean experiment playbook, and like any lean experiment it only produces a usable signal if two things are already in place: a value proposition specific enough that a stranger can react to it in ten seconds, and a source of traffic that resembles your actual target customer. Skip either one and the test still runs — it just measures something other than demand.

A vague value proposition collects vague reactions. "A better way to manage your finances" describes half the App Store; nobody clicks a button, joins a waitlist, or hands over an email address for a promise that generic. Write the one-sentence pitch the way you'd say it to a specific person with a specific problem: "Turn your freelance receipt photos into a categorized expense report every Friday, automatically" gives a stranger something concrete to say yes or no to. If you can't write that sentence yet, the smoke test isn't ready — go narrow the problem first, then come back.

Traffic quality matters more than traffic volume. Ten visitors from a forum thread full of your exact target customer, complaining about the exact problem you're solving, produce a more honest signal than ten thousand visitors from a general subreddit or your own social following, most of whom are other founders rather than buyers. Friends, family, and existing followers are close to the worst traffic source for a smoke test specifically because they already like you — their goodwill contaminates the one thing you're trying to measure. Traffic source is part of the experiment design, not an afterthought bolted on after the page is built.

Pattern 1: The Fake Door Test

A fake door test adds a button, menu item, or feature link that doesn't work yet inside a product or page people already use, then measures how many people click it. Clicking reveals an honest "coming soon — want to be notified?" message instead of the feature itself.

It's the cheapest, strongest signal on this list, but only if you already have somewhere to embed it: an existing app with active users, a live website with real traffic, a browser extension already installed on people's machines. Without one of those, there's no door for anyone to walk up to, and you'll need a different pattern. Where it applies, setup usually takes an afternoon — add the element, wire up a click event, write the honest message behind it.

Pattern 2: Landing Page + Paid Ads

A landing page smoke test describes the finished product on a single page as if it already exists, then drives targeted traffic — usually paid ads, sometimes a relevant community or newsletter — to see how many visitors convert into a signup, waitlist join, or pre-order.

This is the pattern to reach for when you have no existing product and no existing audience, which is most first-time indie hacker ideas. Build time is usually a day: one headline, three or four bullet points on the outcome (not the feature list), and one clear call to action. Ad spend buys you traffic quality you can target precisely, which matters more here than in any other pattern, because a landing page with the wrong visitors will always underperform regardless of how good the pitch is. If this pattern is the one you need, a landing page experiment that proves demand goes deeper on page structure, traffic budgeting, and reading the result honestly than this overview does.

Pattern 3: The Waitlist Test

A waitlist test is a simpler cousin of the landing page test: instead of pushing straight for a purchase or a deposit, you ask for an email address in exchange for early access, and measure how many people give it.

It's the lowest-friction pattern to run, which is exactly its weakness — an email address costs a visitor almost nothing, so a long waitlist by itself proves less than founders want it to. It earns its keep as a first filter when you already have a cheap channel (a newsletter, a community you're active in, a cold outreach list) and want to gauge interest before spending money on ads, or as a funnel-builder to convert later into a stronger test like a paid pilot.

Pattern 4: The Explainer Video Test

An explainer video test replaces the landing page headline with a short video — sixty to ninety seconds, even a screen-recorded mockup with a voiceover — that walks through the product as if it exists, posted where your target customers already gather, with a call to action linking to a signup page.

Reach for this pattern when the product is hard to explain in a single headline: a workflow with several steps, a technical integration, anything where "show" genuinely beats "tell." Setup runs from half a day to a couple of days depending on production quality, but the video's job is clarity, not polish — a founder narrating over static slides that clearly explains the before-and-after is more persuasive than a slick video that leaves the viewer unsure what the product actually does.

Which Smoke Test Pattern Fits Your Situation

The four patterns trade cost, signal strength, and setup time against each other, and the right choice depends more on what you already have (a product, an audience, a budget) than on which pattern is "best" in the abstract.

Read the table as a decision filter, not a ranking. If you already have a product or an audience, the fake door test gives you the strongest signal for the least effort — run that first. If you're starting from nothing, a landing page with narrowly targeted paid traffic is the closest substitute, because a stranger paying attention long enough to click and hand over an email is nearly as costly to fake as clicking a real button inside a real product.

Worked Example: A Landing Page Smoke Test From Setup to Decision

Priya, an indie hacker who consults part-time, keeps hearing the same complaint from other freelancers: reconstructing a quarterly expense report from a shoebox of receipt photos eats an entire Sunday. Her idea, ReceiptPilot, would let someone forward receipt photos by email or text and get back a categorized report every Friday. She has no existing product and no existing audience, so a fake door test is off the table — she runs a landing page test instead.

Setup. She writes the one-sentence pitch from the prerequisites above, builds a single page with that headline, three bullet points describing the Friday report, and one call to action: "Get notified when ReceiptPilot launches" — an email field, nothing more. Total build time is one evening, using an off-the-shelf page builder rather than custom code. She sets a modest budget on a narrowly targeted ad campaign aimed at people who follow freelance-tax and bookkeeping communities, instead of a broad "small business" audience.

Threshold, decided in advance. Before the ads go live, Priya writes down what would count as a "go": a conversion rate meaningfully above what a generic landing page typically manages, and enough total signups that a follow-up test would have someone to run it on. She also writes down what "no" looks like, so she isn't tempted to move the goalposts once real numbers show up.

Running it. The campaign runs for eight days and reaches several hundred targeted visitors. A double-digit number of them submit their email — a conversion rate that clears the threshold she set before she saw any data. More useful than the raw number: several of those signups reply to the automatic confirmation email, unprompted, to ask when it launches or to describe their own receipt-chaos story — evidence closer to what the complete guide to startup idea validation treats as a genuine demand signal than a raw signup count ever is.

Decision. Priya's page cleared the threshold she set before she saw any data, so the honest read is "go" on more validation, not "go" on building. She moves next to a small paid pilot with a handful of the people who replied, to test willingness to pay before writing any code. If the campaign had landed well under her threshold, with nobody replying to the confirmation email unprompted, the honest read would have been "no" — and the fastest thing to do with that answer is kill the idea and move to the next one, not rerun the same test hoping for a different number.

Mistakes: Vanity Clicks and Ethical Lines

Counting raw traffic instead of costed action. Pageviews and impressions cost a visitor nothing to generate; they measure whether your ad targeting worked, not whether anyone wants the product. Count the action that costs the visitor something — a click that leads to a real form, an email handed over, a deposit paid.

Sending your own network instead of cold traffic. Friends, former colleagues, and existing followers already like you, which means their conversion rate measures goodwill, not demand. A smoke test aimed at your own network will almost always look better than one aimed at strangers, and that gap is the whole problem.

Setting no threshold, then rationalizing whatever number shows up. Without a number written down in advance, any result can be read as encouraging — a low conversion rate becomes "early days," a small waitlist becomes "quality over quantity." Decide what counts as a "no" before you see the data, exactly the way Priya did above.

Charging real money with no intention of building or refunding. Taking a non-refundable pre-order for something you already know you won't ship, or charging a card behind a "Buy Now" button that will never process a real purchase, crosses from testing into deception. If a payment is part of the test, the offer has to be real: refund it, or build it.

Manufacturing fake urgency or fake scarcity. "Only 3 spots left" when there's no product and no spots is a lie that inflates your signup rate and misleads the person acting on it — it poisons the very signal you're trying to read honestly.

Reading too little traffic as a verdict. A dozen visitors is noise, not a result, in either direction. Decide your minimum sample size before you launch, alongside your threshold, so a slow first day doesn't get over-read as failure or a lucky first hour doesn't get over-read as success.

Key Takeaways

• A smoke test measures a costed action, not an opinion. A click that goes somewhere, an email address, or a deposit tells you something; "I'd definitely use that" tells you nothing.
• Match the pattern to what you already have. A fake door test needs an existing product or audience; a landing page with paid ads is the substitute when you're starting from nothing.
• Value proposition and traffic quality are prerequisites, not optional polish. A vague pitch and friends-and-family traffic both quietly turn the test into theater.
• Decide your threshold before you see any data. Writing down what counts as "go" and "no" in advance is what keeps you from rationalizing a disappointing number after the fact.
• A waitlist number matters less than what happens next. Signups are the cheapest thing to collect; what people do when you ask them for something real is the actual signal.
• Real money and real urgency claims have to be real. A pre-order you won't refund or a fake "spots left" counter isn't an aggressive test — it's a lie that corrupts the result and misleads the person acting on it.
• Small samples produce noise, not verdicts. Set a minimum sample size in advance so a slow first day or a lucky first hour doesn't get mistaken for the real answer.

Frequently Asked Questions

What is a smoke test in startup validation?

A smoke test is an experiment that measures demand for a product before it's built, using a fake landing page, a disabled feature, a waitlist, or a short video instead of working software. It works by showing people a real offer and counting who takes a costly next step — a click, an email, a deposit — rather than who reacts positively.

Is a fake door test ethical, or is it deceptive?

A fake door test is ethical when the message behind the button is honest — "this doesn't exist yet, want to be notified when it does" — because nobody is misled about what they're getting. It becomes deceptive the moment you take a real payment for something you don't intend to build or refund, or fabricate urgency ("only 3 spots left") that isn't true. The line is whether the person clicking learns the truth immediately, not eventually.

How much traffic do I need before a smoke test result means anything?

There's no fixed number that works for every idea, because it depends on your conversion threshold and how much noise you can tolerate in the result. What matters more than hitting a specific visitor count is deciding your minimum sample size and your target conversion rate before you launch the test, so a handful of early clicks in either direction doesn't get mistaken for a verdict.

What's the difference between a smoke test and a landing page test?

A landing page test is one specific pattern within the broader category of smoke tests — the other patterns are the fake door test, the waitlist test, and the explainer video test. All four share the same underlying logic: show people an offer for something that doesn't exist yet, and measure a costed action instead of an opinion. Which pattern to use depends on whether you already have a product or audience, as the comparison above lays out.