Free UX audit for AI-built sites and apps
You built a site with Lovable, Bolt, v0, Replit, base44, Google AI Studio or Cursor. It works. Now check that people can use it. Paste the URL and get a full UX report in about a minute. No sign-up, no payment.
One public page per audit. Scored against our published checklist of 54 checks.
Common UX problems in AI-generated sites
Many of the sites people run through our audit were built with AI app builders. The builders are good at producing a page that looks finished. They are less good at the details that make a stranger trust the page and act on it. These are the problems to look for, and the checks that catch them.
Generic hero copy
AI builders write a headline that could sit on any site: "Transform your workflow", "The future of X". A visitor cannot tell what you do or who it is for.
Checks: A value proposition in the first screen, H1 says something specific, Enough copy to explain the offer, Average sentence under 22 words
The audit checks that a specific heading and a value proposition are in the first screen. It cannot judge if your wording is generic, so read your headline out loud and ask if it names the product and the customer.
Missing trust signals
Generated sites often ship with no privacy policy, no contact details, no about page and footer links that go nowhere. Strangers do not buy from a page with no owner.
Checks: Privacy policy linked, Terms or legal page linked, A way to contact a human, About / who-we-are page linked, Copyright / owner line present, Social proof on the page
Accessibility gaps
Icon buttons with no label, inputs with no label, images with no alt text and headings picked for size instead of order. These block screen reader and keyboard users.
Checks: No critical accessibility violations, No serious accessibility violations, Images have alt text, Form inputs have labels, Buttons have an accessible name, Heading levels do not skip
Small tap targets on mobile
Designs are usually previewed on a desktop screen. On a phone, icon links are too small to tap, text is too small to read and wide elements make the page scroll sideways.
Checks: Tap targets are at least 44×44px on mobile, No text under 12px on mobile, No horizontal scrolling on mobile, Responsive viewport meta tag
No clear single call to action
Three or four buttons compete in the first screen, or the only button says "Get started" with no hint of what happens next.
Checks: A call to action above the fold (desktop), A call to action above the fold (mobile), Between one and five calls to action above the fold, CTA wording is specific, not "Submit", A way to convert on the page
Placeholder content
Lorem ipsum, "Your Company", example.com links, stock testimonials and buttons that point to "#" survive into production more often than people expect.
Checks: No empty links, No "click here" style link text, No broken or failing requests, Open Graph tags for sharing, Favicon present
The audit does not search for placeholder text by name. It flags the signs around it: empty links, failing requests such as missing images, missing share tags and a missing favicon. Search your pages for "lorem", "example.com" and "Your Company" yourself.
Slow, heavy pages
A client-rendered app can ship a large JavaScript bundle, so the first screen stays blank while it loads. Console errors from leftover code are common too.
Checks: Largest Contentful Paint under 2.5s, First screen paints in under 2s, Page finishes loading in under 5s, Cumulative Layout Shift under 0.1, No JavaScript errors in the console
The audit measures the effect, such as load time and Largest Contentful Paint. It does not report bundle sizes.
What the audit checks
54 checks in four categories, with fixed weights. The same checklist runs on every site, whatever built it.
Usability 35%
Can people actually use the page — on a phone, with a keyboard, with a screen reader, and without waiting?
22 checks
Copy 20%
Does the writing tell a visitor what this is, in plain language, fast?
10 checks
Trust 20%
Does the page give a stranger reasons to believe you are real and safe to deal with?
12 checks
Conversion 25%
Is there one obvious next step, and is it easy to take?
10 checks
How it works
- Chromium loads your page twice: a 1440×900 desktop and a 390×844 phone viewport.
- It scrolls the page while recording, takes screenshots, runs axe-core, and reads the DOM, console, network and Performance API.
- Those measurements are scored against the checklist. You get the report on screen with a fix for each check.
It is an automated pass over one URL, not a human review. It will not catch problems that only appear after login or deep in a journey.
Run the audit on your AI-built site
Frequently asked questions
Is the audit really free?
Yes. You get the full report on screen. There is no sign-up and no payment.
Does it work on sites built with Lovable, Bolt, v0, Replit, base44, Google AI Studio or Cursor?
It works on any public URL, whatever built it. The audit opens the page in a real Chromium browser, so pages that render with JavaScript are checked as visitors see them. The page must be public: the audit does not log in, and private, local and internal addresses are refused.
Does it check my whole app?
No. It audits one page: the URL you give it. Start with your home page or landing page, then run it again on your sign-up or pricing page. It does not fill in forms or follow a journey through the app.
Is the score made up by an AI model?
No. Every page is scored against the same published checklist of 54 checks with fixed weights. The same measurements give the same score.
How do I fix what it finds?
Every check on the published checklist comes with a fix. If you still build with an AI tool, you can paste the failed check and its fix into your prompt and ask the builder to apply it, then run the audit again.
How long does it take?
Usually under a minute. Audits run one at a time, so you may wait in a short queue.
More on the tool: how the free UX audit compares and the best AI website audit tools.