Why I built UXScore
Most UX audits end up in Figma comments or 40-page PDFs that nobody reads. UXScore makes the feedback actionable from the first second.
UX feedback shouldn't need a translator.
Every team has a designer who spots the problems. The heading hierarchy is broken. The contrast ratio fails accessibility standards. The main call-to-action is competing with three other elements for attention. They write it up carefully, thoroughly. Then it goes into a Notion doc or a Jira ticket, and silently dies.
The problem isn't that people don't care about UX. It's that the feedback isn't in a language they can act on. Product managers want impact, not heuristics. Engineers want specific, testable changes. Marketers want to know if the page converts. When a UX audit comes back as a list of technical violations, it gets filed, not fixed.
UXScore was built to close that gap. It takes any URL, understands what the page is actually trying to do, and returns a score with plain-language findings, the kind you can read in 90 seconds and immediately know what to fix first.
Four steps from URL to clarity.
Capture the page
Using Playwright, UXScore takes a full-page screenshot and extracts the complete HTML structure of any URL. No browser plugins, no copy-paste, just a link.
Understand its context
Claude AI reads the page to infer its industry, target audience, and primary conversion goal. A fintech landing page and a portfolio site get scored differently, because they should be.
Score against a rubric
25 checks across WCAG accessibility, UX patterns, and content quality all weighted by relevance to the page's detected goal. The score reflects how well the page serves its actual purpose.
Plain-language report
Every finding is translated into language anyone can act on. Not "fix your ARIA labels" but exactly which element is missing, why it matters to this page's audience, and how to fix it in one step.
The stack.
I built UXScore end to end, the product thinking, the design system, and the full stack behind it. It started from a simple frustration: good UX feedback too often arrives in language only specialists understand, so it gets filed instead of fixed. I wanted a tool that looks at any page the way a careful designer would, then explains what it sees in plain words anyone on a team can act on.
Every decision here: scoring a page against its actual goal, writing findings without jargon, keeping the interface calm and readable,comes back to that one idea: feedback is only useful if someone can act on it.