360° Web Audit Hub
Unifies 4 audit types under a single interface. Includes a GEO module tested against real LLM queries.
- Status
- prototype
- Year
- 2025
- Stack
- Lighthouse, Accessibility (RGAA), Playwright, Python
Why this project
A thorough website audit today cuts across several tools that don’t talk to each other: Lighthouse for performance, an accessibility scanner, a manual extraction of the tagging plan, and increasingly a close read of how well an AI agent can understand the page. Unifying these 4 audit types under one interface, with comparative scoring against competitors, saves real time at the scoping stage.
Use cases
- Tracking Core Web Vitals (LCP, CLS, INP) across a panel of key pages
- Flagging priority accessibility non-conformities (RGAA)
- Mapping the tags and variables firing on each page (TMS)
- A GEO module: tests run against real LLM queries, not simulated ones
- Comparative benchmark against up to 15 competitors, with a consolidated score
How it works
Each audited page runs through a Playwright pipeline that executes Lighthouse, checks accessibility against RGAA criteria, captures network requests to spot outgoing tags, and submits the page to real LLM queries to assess its readability by an agent (semantic structure, structured data, content accessible without JavaScript). The same pipeline runs across a panel of competitors to produce a comparative score, not just an isolated audit.
What I took away
Web performance, RGAA accessibility and GEO readability are more connected than they seem: a poorly structured page penalises users, classic crawlers and AI agents alike. Benchmarking against 15 competitors rather than auditing a single site also changes the conversation with a client: the score immediately situates the effort required.