about / method
How the audit is built
One page, read once: robots, the HTML, a bounded extraction and one optional browser capture. Everything else is arithmetic on top of that.
how a run moves
robotsThe origin’s robots.txt is read first and answered honestly: if it declines GeoNewBot, the run stops there and says so rather than fetching anyway.
fetchOne request to the submitted URL, keeping the status, the headers, the redirect chain and the HTML body exactly as the server sent it.
extractA bounded extractor, isolated from everything else, reads main text, headings, links, images, meta declarations and JSON-LD out of that source HTML.
scoreThe catalog runs across six weighted categories, each row producing pass, warn, fail, unknown or not-applicable with its evidence attached.
labOne optional desktop capture compares the navigation response with the rendered DOM. Those observations are reported, never scored.
reportFixes are ranked by measured severity and effort, with failing access blockers printed beside the score so a good average cannot hide one.
limits we state up front
- One page is read. It says nothing about the rest of the site.
- A pass means a signal was present, not that any assistant cited the page.
- Thresholds are our product behaviour, not ranking requirements published by any vendor.
- Unknown means unmeasured. It is never counted as a pass or as a warning.
run a free auditOne URL. No signup.