Free landing page analyzer
Paste the address of a landing page. We load it on a phone and a desktop the way a visitor would, check what they see first, and give you prioritised fixes with the evidence behind each one — plus the A/B tests worth running. No signup, and nothing on your page is changed.
Any public page — yours or a competitor’s. We load it like a visitor would; nothing is clicked, submitted or changed.
What the analyzer checks
Eight areas, each a handful of checks. Every finding says what it saw, so you can open the page and see it too.
- Message
- Whether there is one clear headline, whether it says an outcome or leans on stock phrases, and whether a phone visitor sees it without scrolling
- Call to action
- Whether the main button is on the first screen, says what happens, and is not competing with four others
- Trust
- Testimonials, logos, ratings and customer counts; HTTPS; privacy and contact links
- Mobile
- Sideways scrolling, tap targets too small to hit, text too small to read
- Speed and stability
- Largest Contentful Paint and layout shift on a phone load, page weight, errors while loading
- Search and sharing
- Title, description, the preview image a shared link shows, and whether search engines are shut out
- Accessibility
- Alt text, form labels, page language, controls with no name
- Measurement
- Whether any analytics is installed — without it, no change can be proven to work
How the score works
Each check passes, earns half credit as a warning, or fails, and checks that matter more to a visitor — the headline, the button, the first phone screen, proof, load time — count for more. The score is the share of the available credit your page earned.
There is no AI in it. The same page in the same state always gets the same score, so you can fix something, run it again, and see the number move for a reason you can name. A check that could not run is marked “not measured” and left out, never quietly counted as a pass.
The score is a way to compare a page with itself over time, or with a competitor’s. It is not a conversion rate: a page can score 90 and still sell the wrong thing to the wrong people.
From findings to A/B tests
Every test idea comes from a finding on your page, and uses your page’s own words as the control — your actual headline, your actual button. An idea with nothing on the page behind it is not in the list.
Each one is written as a hypothesis with a single metric, so it can be run as a real test rather than a redesign. If the page has no analytics, the report says so first: without it, no test can tell you which version won.
Building the variant is usually the slow part. Refinar can make any of these changes in your repository as a live preview, ready to put behind your testing tool, without waiting for a sprint.
What it cannot see
Worth knowing before you trust a score. An outside check sees what a first-time visitor sees, and no more.
- Your conversion rate. The analyzer sees the page, not the people on it — only your analytics or a real test can say whether a change worked.
- Pages behind a login, a password or a paywall. It sees what a first-time visitor sees, and nothing more.
- Whether the page matches the ad, email or post that sends people to it — often the biggest lever of all.
- Anything after the click: the sign-up form's next step, checkout, onboarding.
Questions about the analyzer
Is it really free? Do I need an account?
Yes, and no. Paste an address and you get the whole report — the score, the findings, the screenshots and the test ideas. An account only matters if you want Refinar to make the changes in your code.
Does it use AI to judge my page?
No. Every finding comes from a fixed check against what the page actually contains or how it actually rendered, so the same page always gets the same score. Most analyzers pass your HTML to a language model and return its opinion, which changes each time you ask — a problem for a number you want to track or share with your team.
Can I analyse a competitor's landing page?
Yes. It works on any public page and reads only what any visitor could see. Comparing your first screen with a competitor's on a phone is one of the more useful things to do with it.
What is a good score?
Above 85 means the fundamentals are in place and the next gains come from testing your offer and message. Between 60 and 85 is typical: a few fixes with obvious payoff. Below 60 usually means something basic — the button off the first phone screen, no proof, a slow load — is costing you visitors before your copy gets a chance.
How is this different from PageSpeed Insights or Lighthouse?
Those measure speed and technical health in depth. This checks what makes a landing page convert — the headline, the call to action, proof, the first phone screen — and includes the speed numbers that matter to a visitor. Use PageSpeed Insights when a developer is tuning performance; use this when you are deciding what to change on the page.
Why does it sometimes say a check was not measured?
Layout checks need a real browser, and those runs are limited per visitor. When one does not run, the checks that need it are marked “not measured” and left out of the score — never counted as a pass or a fail.
How do I get the fixes made?
Copy the report into a ticket and hand it to whoever edits the site. Or connect the repository behind the page to Refinar: it maps each finding to the file responsible, makes the change in a live preview, and opens a pull request for you to approve.