Landing-Page Test — The Product Playbook

One page that sells your product as if it already exists, with a sign-up or buy button that records who meant it. You drive real traffic to it and count who acts. No product behind the button yet. Just the page and the promise.

It's the cheapest way to ask a stranger to vote with a click before you build a line of code.

The question it answers

One question, and it's the one that should scare you most about a brand-new idea: does anyone actually want this?

That's desirability. Whether anyone wants the thing at all. Anyone can get excited about building. A stranger's click costs them something, and that's what this page buys you.

A survey can't do this. A survey collects opinions, and opinions are free. This page makes people do a thing instead of say a thing, and the doing costs them something real: a click, an email, thirty seconds of attention. When someone acts, you've learned something. When someone says yes to a survey, you've learned they're polite.

It tests your messaging too, since the words on the page are the pitch. If nobody clicks, the story might be the problem, not the product. But the question it's built for is desirability.

How to run it cheaply

The minimum version is genuinely one page and a week.

Build one page that sells the thing. Name the problem in the reader's words, show what the product would do, and put one button front and center: "Reserve yours," "Get early access," "Buy now." No backend needed. The button can collect an email or just record the click.
Write your success number down first, before launch, so it can't bend afterward. For a brand-new concept you're putting in front of cold strangers, a sign-up conversion of a few percent is a reasonable starting bar, two to five out of every hundred who land hitting the button. Published landing-page medians actually run higher (around 6% across industries, closer to 4% for software, per Unbounce's benchmark data), but those come from established pages with warmer traffic, so keep your bar modest for a cold test like this one. Pair it with a sign of real reading, like time on page above 30 seconds, so a bounce doesn't count as interest. Or set a flat target: 100 reservations in week one.
Point real traffic at it. A small ad budget for a week is enough. The traffic has to be strangers, not your friends and not your email list, because the whole point is whether total strangers click the button.
Count who acts, then decide against the number you set. Hit the bar, and you proceed. Miss it but the messaging clearly worked, and you refine the page and rerun. Nobody acted, and you kill it.

Treat any benchmark as a starting point, not a promise. Your real target gets set for your audience, in your context, before launch.

A worked example

The company and numbers below are invented to make the mechanics concrete. The method is real and the benchmark is real and credited. Nothing here is a result anyone is claiming. It's a recipe, shown running.

Imagine a meal-kit service for people who work twelve-hour hospital shifts: pre-portioned, heat-in-ten-minutes, dietitian-approved. Before you sign a single supplier, you build one page.

It names the problem in their words. "You get home at 9pm and there's nothing to eat that won't wreck you." It shows three sample meals. It puts a "Reserve your first box" button right up front.

You write the success criterion down first, so it can't move on you later. Sign-up conversion of 3% or better, plus average time on page above 30 seconds, so you know people read it instead of bouncing. Then you run a small ad budget to the page for a week.

Picture 1,800 strangers landing on it. 63 hit Reserve. That's 3.5%, inside the band, and average time on page is 48 seconds. You cleared the bar you set. People who'd never heard of you read the pitch and reached for the door.

That's a real demand signal, bought for the price of one page and a week of ad spend. Not a quarter spent building a supply chain nobody wanted.

When to use it, and when not

Lean experiments come in a ladder, from the cheapest and lowest-commitment tests up to ones that cost real time and money. Use the landing-page test when you have a value proposition you can describe and you want a market's verdict before you build. It's a bottom-rung experiment: cheap, fast, no engineering, low commitment. Start here.

Reach for something else when the shape of the question changes. The siblings below all live in the same lean-experiments catalog:

You already have live traffic and you're testing one new feature, not a whole product. That's the fake-door test, a button for a feature that doesn't exist yet, dropped inside the product you already run.
The product is impossible to picture from words alone. When seeing it move is what makes someone want it, you want the smoke test, the explainer-video version of this idea scaled up to the whole product. Dropbox famously posted a short video of the product working before they had built it, then measured the sign-ups it drove.
You want to test the words before the page is worth building. Run a social-media teaser or an online ad test first, then let the winning angle go on the landing page you build next.
A click feels too cheap to trust. It is cheap. When you need the interest to survive real money, climb to a pre-sell: a page that actually charges or takes a real card up front, not just a click. That's a higher rung for when this one has already come back positive.

The honest version cuts both ways. If almost nobody acts, you just saved yourself a quarter. A landing page nobody clicks is one of the cheapest "no" answers you'll ever buy, and a no you bought for a week of ad spend beats a yes you guessed at for six months.

The taxonomy underneath this experiment, and the deeper catalog of the rest, 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.