How to validate a startup idea with a landing page before you build
A landing page can't prove a business works, but it can tell you cheaply whether strangers care enough to leave an email address. The test only means something if you fix the audience, the promise, the traffic and the pass mark before you start. Here is the sequence, with what Anticode does and what you run yourself.
What this kind of test can and can't tell you
A smoke test measures stated interest: people read a promise and asked to hear more. That's weaker than a payment and stronger than a friend saying it sounds good. It answers "does this framing make this audience act?", not "will they pay what I need, every month, for two years". Treat a pass as permission to do the next, more expensive test, such as ten customer calls or a pre-order.
It also measures the page as much as the idea. A muddled headline can fail a good idea. That's why the steps below keep the page deliberately plain and change one thing at a time.
The test, step by step
- Write the hypothesis in one line. "Independent physiotherapists will join a waitlist for a tool that writes appointment reminders in their own voice." One audience, one outcome.
- Set the pass mark before anything is live. Decide the number of visitors you'll send, the sign-up rate that means go, and the rate that means stop. Write both down, with the date. The band in between means change the promise and run again.
- Build the page. In Anticode, describe the product and audience in a sentence or two. The Architect asks up to three questions where the answer changes the page, then shows a plan. Ask for a landing page with the newsletter sign-up as the main action. The design engine publishes plain HTML and CSS with no scripts and no tracking.
- Cut it down. Remove any section that makes a claim you can't stand behind yet: invented feature lists, made-up testimonials, a pricing grid for tiers that don't exist. A headline, two or three lines on the problem, the form, and who you are.
- Send sign-ups somewhere you'll use. The newsletter form asks for an email only. Each sign-up is saved in the dashboard; a connector can add it to Kit or Mailchimp (Mailchimp uses double opt-in) or a Google Sheet.
- Publish and send known traffic. Use one channel per test: a small ad budget, posts in two communities where the audience gathers, or direct outreach to a list. Note what you sent and when.
- Read the numbers. Compare sign-ups in the dashboard with page views. Anticode's analytics count views per day by source, without cookies, so you can see which channel brought the visitors.
- Apply your rule. Go, stop, or change one element and rerun. Don't move the pass mark after seeing the result.
Choosing the pass mark
There's no universal sign-up rate that means yes, because cold ad traffic, warm community traffic and people you messaged directly behave very differently. Set the mark from your own economics: if you need a certain number of paying customers and expect only a fraction of sign-ups to pay, work backwards to the sign-ups you need from the traffic you can afford. Also set a minimum sample. A handful of visitors tells you almost nothing; decide the number of visitors you'll wait for before reading the result at all.
Running a second round
If the result lands in the middle band, change one thing: the headline, the audience you send, or the specific outcome you promise. Ask for the change in a sentence ("rewrite the headline for clinic owners rather than solo therapists") and the page updates; the earlier version stays in history for comparison. Anticode has no A/B testing, so rounds run one after the other, each with its own traffic and dates. Keep a simple log of each round's dates, promise, channel, views and sign-ups.
What can mislead you
- Mixed traffic. Friends who sign up to be kind inflate the rate. Keep personal networks out of the test, or count them separately.
- Bot and crawler visits. Anticode counts AI and search crawlers separately from page views, but check sources for anything that looks odd before you calculate a rate.
- The free plan's allowance. On the free plan a monthly allowance of submissions is visible (100 by default); the rest are held until you upgrade. For a large test, check this against your traffic first.
- Silence afterwards. Sign-ups go cold. Write to them within a day or two from your email tool; Anticode doesn't email your list.
Where Anticode's part ends
Anticode builds and publishes the page, stores and forwards sign-ups and counts views by source. It doesn't run ads, split traffic between variants, record what visitors click, or email the people who sign up. If you want to test willingness to pay, the next step is a pre-order page with a payment link from your own provider, or an honest fake door test. For the page itself, see the waitlist page guide.
Questions people ask
How long should the test run?
Until you reach the visitor count you decided in advance, not a calendar date. If a channel can't deliver that many visitors in a few weeks, pick a different channel.
Should I ask for more than an email address?
Not in a first test. Every extra field lowers the rate you're trying to measure. Anticode's newsletter form asks for an email only; ask follow-up questions in your first email.
Do I need a custom domain to validate a startup idea with a landing page?
No. A free anticode.dev address works, though some visitors trust a real domain more. Custom domains are on paid plans.