What Is a Smoke Test in Startup Validation?
A smoke test is a lightweight validation experiment that measures real demand for a product before you build it. You put the offer in front of real people — usually with an ad pointing to a simple landing page — and count a concrete action like a click, sign-up, or pre-order. That action is your demand signal.
Quick Answer: A smoke test shows people a real offer for a product that doesn't exist yet and measures whether they take a real action to get it — a click, an email sign-up, a pre-order. The action, not the stated opinion, is the evidence. Where there's smoke (real interest), there's often fire (a real market).
The name borrows the old idea that where there's smoke, there's fire. If a fake offer draws genuine clicks and sign-ups, there is probably a real fire — demand — worth building for. If it draws silence, you learned that in days instead of finding out after months of engineering.
One quick disambiguation. Engineers also say "smoke test" to mean a fast check that a new build runs without catching fire before deeper testing starts. That is a different thing entirely. In startup validation, a smoke test is a demand test, not a software check.
The idea sits at the heart of Alberto Savoia's The Right It. Savoia argues that most new products fail not because they are built badly but because nobody wanted them — and that the only reliable defense is to collect your own data from people with skin in the game, rather than trusting opinions. A smoke test is exactly that: a small, real market signal you gather yourself. It is a close cousin of the fake door test, and it usually rides on a landing page.
How a Smoke Test Works: Ad, Page, and Signal
A smoke test works by faking the front of a product and measuring who tries to walk through it. You don't build the product; you build the believable promise of it and watch who reaches for the handle.
Every smoke test has three moving parts:
- A source of real traffic. Usually a small paid ad, but a post in a relevant community, a cold email, or a link in a newsletter all work. The point is that strangers who match your target customer meet the offer without you hovering over their shoulder.
- A page that states the offer. Most often a single landing page MVP that describes the product as if it already exists — the headline, the core promise, and a price. It doesn't need to be polished; it needs to be believable.
- A measurable action — the signal. The one thing you count: a click deeper in, an email on a waitlist, a pre-order, a "notify me" tap. Choose that action before you launch, so you can't quietly move the goalposts afterward.
The action is the whole point. A survey where someone says "yeah, I'd use that" costs them nothing, so it proves almost nothing. A sign-up costs an email and a flicker of real intent; a pre-order costs money. The more the action costs the person, the stronger the signal — this is what Savoia means by skin in the game. You want evidence that people will do something, not merely say something.
A Concrete Smoke Test Example
Here is a smoke test you could run this week. Suppose you suspect busy parents of kids with food allergies would pay for an app that generates safe weekly meal plans. Instead of building it, you test the demand first:
- The offer: a one-page site headlined "Allergy-safe meal plans, done for you," three bullet points, a price of $9/month, and a single button — Start my plan.
- The traffic: a small ad budget aimed at allergy-parent groups and related search terms.
- The signal: taps on Start my plan that land on a "Join the early list — we launch soon" email capture.
Say the ad sends 400 people to the page and 48 tap Start my plan and leave an email. Those figures are illustrative — your real numbers are the experiment. What matters is that 48 people took a slightly costly action toward a product that does not exist. Now compare that with the same budget behind a page headlined "Meal planning made easy" that collects four emails. Same effort, very different fire — and you learned which promise resonates before writing a line of code.
One rule keeps this honest: don't take money for something you can't deliver. A waitlist or a "notify me at launch" measures real intent without misleading anyone.
Smoke Test vs Fake Door vs Landing Page Test
These three terms overlap so heavily that founders use them interchangeably, but each answers a slightly different question. The table below lines up the demand tests you'll hear about most.
| Dimension | Smoke test | Fake door test | Landing page MVP |
|---|---|---|---|
| Core idea | Advertise an offer for an unbuilt product and measure who acts | Advertise a specific feature that isn't built and measure who tries it | A single page that describes the product and captures interest |
| Where the offer lives | Usually an ad plus a standalone page | Usually a button or menu item inside an existing product | The page itself is the test |
| What you measure | Clicks, sign-ups, or pre-orders on the offer | Clicks on the missing feature, then the "coming soon" drop-off | Sign-ups, waitlist joins, or clicks from the page |
| Best question it answers | "Does anyone want this product at all?" | "Do my existing users want this specific thing?" | "Can this concept turn cold interest into intent?" |
Takeaway: The lines blur on purpose. A smoke test often runs on a landing page, and a fake door test is really a smoke test scoped down to a single feature. Don't agonize over the label — ask what you're trying to learn: whole-product demand (smoke test), feature-level demand (fake door), or whether a particular page converts (landing page test).
When a Smoke Test Is the Right First Experiment
A smoke test is the right first experiment when your biggest unknown is whether anyone wants the thing at all, and when you can describe the offer clearly enough to put a price and a button on it.
Reach for a smoke test when:
- You can state the value in one sentence. If you can't write the headline yet, you can't test the demand yet — run problem interviews first.
- Your risk is "no one wants it," not "can we build it." Smoke tests retire demand risk cheaply. They say nothing about technical feasibility or whether the product delivers on its promise.
- You'd otherwise build for weeks on a hunch. The trade is a few days and a small ad budget against months of engineering.
Skip it as a first move when:
- You need the qualitative why, not a yes or no. A smoke test tells you that people acted, rarely why — pair it with a few conversations.
- Your audience can't be reached cheaply. Very niche B2B buyers may need direct outreach rather than an ad.
Think of a smoke test as the demand gate at the front of validation. It's rarely the only experiment you run — a passing test usually leads to interviews and a rougher build — but it's often the cheapest one to run first. Our complete guide to startup idea validation shows where the smoke test sits in the full sequence, from problem discovery to a build decision. If you'd rather keep these experiments and their results in one view instead of scattered across docs, that's the kind of thing Edmired is built for.
Key Takeaways
- A smoke test measures demand before you build. You advertise an offer for a product that doesn't exist and count a real action — a click, sign-up, or pre-order — as the evidence.
- The action is the signal, not the opinion. A survey "yes" is cheap talk; a sign-up or pre-order carries skin in the game and tells you far more.
- It has three parts: a source of real traffic, a page that states the offer, and one measurable action you pick before launching.
- It's rooted in pretotyping. Alberto Savoia's The Right It argues most products fail because nobody wanted them, so you collect your own data instead of trusting opinions.
- Don't confuse it with the software meaning. In engineering, a smoke test checks that a build runs; in validation, it tests market demand.
- Fake door and landing page tests are close relatives. A fake door scopes the test to one feature; a landing page is often where a smoke test runs.
- Use it when demand is your biggest risk — and pair it with conversations whenever you need to understand why people did or didn't act.
Frequently Asked Questions
Is a smoke test the same as an MVP?
No. An MVP is the smallest version of a product you actually build to start learning from real use. A smoke test builds nothing — it measures demand for a product that doesn't exist yet. A passing smoke test is often what justifies investing in an MVP in the first place, so the two sit at different stages of the same journey.
What counts as a passing smoke test?
There's no universal number. A pass is hitting the action rate you committed to before launching, ideally against a control version. Set that threshold from your own economics — what a customer is worth and what your traffic costs — not from a benchmark you read online. The discipline is deciding what "good" looks like in advance, so you can't rationalize a weak result afterward.
How much does it cost to run a smoke test?
Far less than building. The core ingredients are a simple one-page site and a small amount of traffic, so most smoke tests cost a modest ad budget and a weekend of setup rather than months of engineering time. That asymmetry — a few days against a potential half-year build — is the entire reason the technique exists.
Does a smoke test work without paid ads?
Yes. Any source of real, unbiased traffic works: a post in a community your customers already gather in, a newsletter mention, or targeted cold outreach. The requirements are that the people match your target customer and that they aren't friends being polite. Paid ads are simply the fastest way to reach strangers at a volume that produces a readable signal.