Smoke Test (Explainer Video)

A short video or simple page that shows the whole product working as if it already existed, published so you can count who signs up off the demonstration alone. Nobody's built the thing yet. You build the show of the thing, point real traffic at it, and watch who reaches for something that doesn't exist.

Think of it as a fake-door test scaled up, where a fake-door test is a single fake button for a feature you haven't built. That smaller test fakes one button inside a product you already have. A smoke test fakes the entire product, before any of it is real, and the signup is the signal: a stranger reaching for something that does not exist yet.

The question it answers

Will anyone reach for this at all, when the only thing they can act on is the demonstration?

Any experiment answers one of four questions: desirability (do they want it), usability (can they use it), feasibility (can you build it), messaging (does the pitch land). The smoke test buys a clean read on desirability, because the only thing anyone can act on is the demonstration, and a side read on messaging, because the video forces you to pick the exact words and framing that sell the thing, and the signup rate tells you whether that pitch worked. It tells you nothing about usability or feasibility. That's not what it's for.

The taxonomy underneath all of this, including the smoke test, comes from David Bland and Alex Osterwalder's Testing Business Ideas, which maps more than forty of these tests by the kind of evidence each one buys you.

How to run it cheaply

The whole experiment is one artifact and one number. Here's the minimum version that still raises real confidence.

1. Show the thing working. A short walkthrough, two or three minutes, of the product doing its core job. Screen recording, a few mocked-up screens, a voiceover. It does not have to be polished. Ugly is okay. It has to be clear.
2. Put a door on it. One page, the video, and a single action that records who meant it. "Join the waitlist." "Get early access." An email field is enough.
3. Point real traffic at it. Strangers, not your friends and not your team, because friends and teammates will sign up to be kind to you and give you a flattering read instead of an honest one. A small ad spend, a post to the audience you believe wants it, a link in the right community.
4. Write the success number down first, before you publish. How many signups, against how many visitors, by what date. The number written down first is the one rule this whole experiment leans on, because a number you set afterward will quietly bend to whatever the result was.

That's it. No engineering. No product. A weekend of work and a week of attention, instead of a quarter spent building a thing nobody asked to see.

A worked example

The classic smoke test is Dropbox. They couldn't easily demo file-syncing that didn't exist yet, so instead of building it they shot a short video walking through the product as if it were real, posted it, and watched the beta waiting list jump overnight. The video was the experiment. The signups were the signal.

I'm not going to re-tell the whole Dropbox run here, because you've already got it. Go read the full telling on the Test & Validate page back in Part 04 when you want the play-by-play. The shape there is the shape here: show the thing working, point real traffic at the demonstration, count who signs up off the demo alone.

It's copyable because the artifact matched the risk. Dropbox's riskiest question wasn't "can we build sync." It was "does anyone outside our heads even want frictionless sync." A demonstration was the cheapest artifact that could answer that, so a demonstration is what they built.

When to reach for it, and when not

Reach for the smoke test when the product is hard to picture from words alone, and seeing it move is the thing that would make someone want it. Software with a new interaction, a workflow nobody's seen, anything where a paragraph undersells it and a demo sells it.

Reach for something else when:

• You already have live traffic and you're only testing one new feature. That's a fake-door test, smaller and cheaper. Don't fake the whole product to learn about one button.
• A page and a promise would carry it without a demo. That's a landing-page test. If the value is obvious in a headline, you don't need to film anything.
• You want to learn which words pull hardest. That's a social-media teaser, three angles against each other. The smoke test reads one story; the teaser races several.
• A signup feels too cheap and you need a harder read. Climb to a pre-sell, where you ask people to actually pay or commit money up front, before the thing exists. Real money is a harder read than a free signup, because a signup is a real signal but it's still a low-cost yes.

Think of these tests as a ladder. They rank from cheap and low-commitment, which buys a fast but weak signal, up to expensive and high-commitment, which buys a slow but strong one. The smoke test sits low on that ladder on purpose: cheap, low-commitment, fast. Start there. Climb only as far as your question forces you to.