How to test demand with a landing page before you build

A landing page tests demand when four things are settled before anyone sees it: the one promise it makes, what the button asks for, the number that counts as a pass, and how many visitors the result needs. Get those right and the page answers a real question — whether strangers will commit to the thing you described. Get them wrong and it measures your mood. Below is how to set all four up alone, with no audience, no ad budget and no calls.

What goes on the page

The page has one job: get a stranger who has the problem to take the action. Everything on it either helps that or gets in the way. Our landing-page generator builds the page in the order below, with the button straight under the promise, where a visitor who already recognizes the problem can act without reading further.

Say what exists. "Not built yet — reserve a place" is honest, and it measures the thing you care about: whether they want the outcome, not whether they were fooled into thinking it is ready.

Show the price if price is the question. A price that is not on the page cannot be tested. A validation plan carries a price hypothesis, and our generator is given it when it writes the page, but nothing forces the price onto the page. Check that yours shows it.

Pick the button by what you need to learn

A validation plan can ask a visitor for one of 4 things, and 3 of them take money. The cheaper the button, the less a result can tell you.

Ask for the strongest commitment you can honestly collect. If you can take a payment — a link, a pre-order, a paid pilot — ask for it. A waitlist is the fallback, not the default. A signup can show that the problem is real and that you reached the people who have it; it can never show that the price works. Will people pay? walks the whole ladder, from praise to payment, and what each step can answer.

Write the pass mark before the page goes live

The pass mark is the part that makes this a test. Before launch, write down the one number you will read, the minimum that counts as a pass, the window it has to happen in, and what you will do in each case. Put the visitor count in as a line of its own. After launch, every number looks like a reason to keep going.

A bar set at the number you expect is missed often, even by a good page. A page that really converts at 2% clears this one about 56% of the time, and gets no pre-order at all only 0.6% of the time. So a zero after the full visitor count is a real answer, and a near miss is mostly luck. That is why the partial row earns one more run, not a verdict.

The number has to be one you can actually read. Our validation plans write each pass mark as a line of text and, where it fits one of 5 metrics — visitors, signups, payments, revenue, replies — as a number with a minimum and a window of at most 90 days. When you report how the test went, the numbered lines are graded against what you reported; a line with no number is not. Report once, when the window ends: if you report more than once, the weakest report is the one that counts, so an early report cannot be replaced by a better one later. A second try is a new plan and a new test. Validating without talking to customers is the whole method: why the bar comes first, and what a pass does and does not prove.

Getting enough visitors without an audience

A page nobody saw has not given you an answer. At the conversion rates our no-audience guide probes, 1% to 5% of visitors paying, one paying customer takes on average between 20 and 100 qualified visitors — people who have the problem, not anyone who clicks. That is why the visitor count goes into the pass mark, and why you plan where those visitors come from before launch rather than after the first quiet week. The channels that need no budget, and the arithmetic behind them, are in that guide.

Count visitors from a source that counts everyone. Analytics that wait for cookie consent report a floor, not a total. Divide a complete count, such as payments, by that floor and the conversion rate comes out too high, by an amount you cannot measure.

Fake door tests, and where they stop being honest

A fake door test is a page, or a button inside a product, for something that does not exist yet. The click is the signal. It is a fair test and a common one, with one line you must not cross: nobody should leave believing they bought something that is not coming.

After the click, say what happened. For a waitlist: "It isn't built yet — you are on the list." For money, it is a pre-order, so say when it will be delivered and how to get a refund, and refund anyone you cannot serve.

Never invent the proof. A signup counter you do not have, a logo you were never paid by, a testimonial from nobody — each makes the test measure the fiction instead of the demand. The instructions our generator writes under forbid all three, and a button with no destination renders disabled with a note rather than pointing anywhere made up.

Reading the result

Read it against what you wrote, in this order.

• Not enough visitors. The window closed before the page reached its visitor count. The idea is untested, not rejected: get the visitors, then run it again with a fresh bar.
• Pass. The bar was met inside the window. For a waitlist, that is the signal to ask the same people for money. For a payment, it is the signal to build the smallest version you promised.
• Partial. Some commitment, below the bar. Change one thing — the price or the promise, not both — and run it once more with a fresh bar. Change two and you will not know which one mattered.
• Fail. Enough visitors, no commitment. Quit, as you wrote you would. It is the cheapest outcome the test can give you, and the reason to run it before the code.

What a landing page test cannot tell you

A clean result covers less than it feels like. Read through the scorecard our checks use, the widest pass a market test can produce evidences 55% of it: that the problem is real, that you can reach the buyer, and, if money moved, that the price works. Our grader never counts a market test as evidence for the other 45%: whether buyers trust a one-person business enough to buy from it, how much support they will need, how exposed you are to a platform or a regulator, and whether the cash comes in steadily enough to last. A pass is permission to build a small version, not a verdict on the business.

Running it with WhittleOS

Our validation plan is this page's method, written for one idea: the promise for the page, the button, a price to test, between 3 and 8 async tasks, outreach drafts screened for any request to get on a call, the pass mark in a form that can be graded, when to quit, and what each outcome would mean. It costs 1 credit.

The landing-page generator writes the page from that plan and hosts it at a public link, or gives you the HTML to put on your own domain. The button goes to your own waitlist form or payment link. The hosted page carries no analytics, so to count visitors, count clicks on the link you share or put the HTML where you can measure it. The generator opens once the idea has had its idea check, so the whole path is 3 credits: the 2 free credits every new account gets cover the check (1 credit) and the plan (1 credit), and the page (1 credit) is the first you pay for.

The short version

• A landing page tests demand only when the promise, the button, the pass mark and the visitor count are all fixed before launch.
• Ask for the strongest commitment you can honestly collect. A signup can show the pain and the reach; only a payment can show the price.
• Write the bar as a number, a minimum and a window, visitors included, and judge nothing until the page has had its visitors.
• A near miss on a bar set at your hope is mostly luck; a zero after the full visitor count is an answer.
• A fake door is fair when the page says what exists and never keeps money for something that is not coming.
• A pass covers 55% of what decides the idea. The rest takes customers.