A landing page test is not a way to find out whether people like your idea. It is a way to find out whether a specific promise, shown to a specific group of people, makes them do a specific thing they would not otherwise have done. That is all it measures. Built properly, it can save you months. Built the usual way (vague headline, email box, traffic from your own contacts), it produces a number you will misread as encouragement and then spend six months building against.
The mechanics matter more than the design. Here is what actually makes the test valid.
What a landing page test really proves
A smoke test measures the gap between attention and action. You put a promise in front of a cold stranger and ask them to take one step that costs them something: time, contact details, a card number, a slot in their calendar. If enough of them take it, you have learned that the promise lands with that audience through that channel.
What it does not prove: that they will pay, that they will pay your price, that the thing you build will actually solve their problem, or that you can reach those people repeatably at a cost you can afford. Those are four separate questions and a landing page answers none of them cleanly.
Treat the page as evidence about your message, not proof of demand. Demand is proven by money changing hands. Everything else is a signal of varying strength, and a signal is only useful if you rate it honestly. If you want the wider frame before you build the page, the broader method is laid out in how to know if people will buy.
Write the one promise before you write the page
Most failed landing page tests fail at the headline. The page describes a category ("bookkeeping made simple") instead of making a promise ("your books closed by the 5th of every month, or you don't pay this month").
A testable promise names who it is for, what changes, and by when. It should be specific enough that a stranger can disagree with it. If nobody could read your headline and think "that is not for me", the headline is too soft to test anything.
There is a second trap here. Some offers are instantly understood, and others need explaining before you can argue you are the right choice, a distinction Entrepreneur made well in a piece on helping customers make sense of their options. If your idea sits in the second category, a cold landing page will underperform regardless of demand, because you are testing comprehension and desire at the same time. For those ideas, conversations beat pages. Start with how to talk to customers before you build instead.
Pick the hardest action you can justify
The strength of your evidence is set almost entirely by what you ask people to do. Ranked from weakest to strongest:
- An email sign-up for "early access". Nearly free to give. Says "mildly curious". This is the one founders over-read.
- A waitlist with a qualifying question or two. Slightly better, because it costs attention and filters out tyre-kickers.
- A booked call on your calendar. Much stronger. They spent real time and put their name to it.
- A pre-order, deposit or paid pilot. Strongest. Money is the only signal that cannot be faked into existence by politeness.
A fake door test sits in the middle of that list: the button says "Buy now" or "Start free trial", and the next screen says the product is not open yet and offers to notify them or book them in. Handled well, it captures purchase intent rather than curiosity. Handled badly, it feels like a bait and switch. Be upfront on the next screen, explain plainly that you are validating before building, and offer something real in return, even if that is just a genuine early slot or a call.
Ask for the most expensive action a stranger could reasonably give you at this stage, then rate the result accordingly.
The traffic source is the test
This is the part people skip, and it is the part that makes or breaks the exercise. A page with no traffic proves nothing at all. Worse, a page with the wrong traffic proves something false, in either direction.
Send it to your LinkedIn network and your former colleagues and you will get warm, flattering numbers from people who are responding to you, not to the offer. Send it to a broad interest audience on paid social and you will get cheap clicks from people who have no problem to solve. Both reads are worthless, but they look identical in an analytics dashboard.
Choose one source where the people arriving already have the problem, and make it a source you could actually scale later:
- A small, well-targeted paid test on the platform where your buyers already search or scroll.
- A relevant community, forum or subreddit, where you contribute properly and link only when it belongs.
- Cold outreach to a tight list, with the page as the destination rather than the pitch.
- A partner or newsletter that already serves your exact audience.
One source at a time. If you run three at once you will not know which promise worked for whom, and the cheapest traffic will drown out the most qualified.
What conversion numbers actually mean at small sample sizes
Here is the uncomfortable arithmetic. Say you get 80 visitors and 4 sign-ups. That is 5 percent, and it feels like a data point. But if two of those four had been distracted and closed the tab, you would have 2.5 percent and a completely different mood. At small numbers, one or two people swing your headline metric by half. You are not measuring demand, you are measuring noise with a decimal point on it.
So do not tune your read to the percentage. Read the numbers like this instead:
- Zero conversions from genuinely qualified traffic is the most informative outcome you can get. It is a real signal, and it is cheap. Change the promise, not the button colour.
- A handful of conversions tells you the promise is not dead. It does not tell you the rate. Treat it as permission to go and talk to those people, not as proof of a market.
- Enough conversions that you are busy replying to people is the only volume at which a rate starts to mean anything, and by then you have better evidence available anyway: the conversations themselves.
The useful output of a small test is names, not percentages. Four people who raised their hand and will take your call are worth more than any conversion rate you could calculate from them. That is where validation actually happens, which is also why our First Customer Validation is built around getting to real buyers rather than optimising a funnel.
The honest read, and the follow-up that matters
Every conversion is a lead on a question, not an answer. Email the people who signed up within a day, while the promise is fresh, and ask two things: what made them click, and what they are doing about the problem today. The answers will tell you whether you found buyers or browsers.
Then hold yourself to one rule: decide in advance what result would make you stop. Write it down before you launch the page. "If I get fewer than X qualified calls booked from Y visitors, I rewrite the promise and test a different audience." Founders who skip this step never get a negative result, because they always find a reason the test was unfair. The warning signs of that pattern are collected in 9 signs your business idea will fail.
A landing page cannot tell you whether your idea is good. It can only tell you whether your sentence is good enough to earn a stranger's next five minutes, and that is worth knowing before you spend five months building.
When a landing page is the wrong tool
If your offer is a service sold to a small number of specific buyers, skip the page. Ten direct conversations will teach you more in a week than a smoke test will in a month, because your total market might be a few hundred people and no traffic source reaches them efficiently. The same is true for anything where the buyer needs to trust a person before they will consider the product.
Landing pages earn their place when the audience is large, reachable through a channel, and the promise can be understood in one screen. Otherwise you are building infrastructure to avoid a conversation.
If you are not sure which camp your idea is in, the free five-minute Idea Validation Score at /validate will tell you whether a smoke test is the right next step or a distraction, and what to do instead if it is. If you would rather talk it through with someone who has read a lot of these tests, book a conversation and we will look at your promise line by line. How the paid work is structured, including what each engagement covers, is on /pricing.
Frequently asked questions
How much traffic do I need for a landing page test to be valid?
Enough that a handful of conversions either way does not change your conclusion, which is usually more than a solo founder can send in a first attempt. Rather than chasing a number, aim for a pool of clearly qualified visitors from one channel and then judge the test on the quality of the people who converted, not the rate. If you cannot get qualified traffic at all, that itself is the finding: your distribution problem is bigger than your product problem.
Is a fake door test dishonest?
It depends entirely on what happens after the click. If the next screen is transparent, explains you are testing before building, and gives the person a genuine option (an early slot, a call, a notification they actually want), most people are fine with it. If it takes payment details for something that does not exist, or implies availability you cannot deliver, that is a different thing, and anything involving taking money before delivery is worth a quick check with a professional in your jurisdiction.
Should I build a landing page or just email people directly?
If you can name fifty specific people who have the problem, email them. Direct outreach gets you conversations, objections and sometimes a first sale, all of which beat a conversion rate. Build the page when you need to reach an audience you cannot list by name, or when you want a single place to send traffic from a channel you plan to scale. The practical playbook for the direct route is in how to get your first customer.