The problem
Schools must publish a long list of information online, and websites drift: a policy link goes stale, a page is republished without its key content. Checking by hand is slow and inconsistent, and a scanner that simply declares a site compliant makes a promise it cannot keep.
What I built
A membership scan that reads a school's accessible pages and linked documents each term. Some checks are pattern detection that a script can do alone. Others need the page read in context, such as whether a complaints procedure is a procedure or a sentence saying one exists. That content review uses AI, and its findings carry a label so the board scrutinises them like any written judgement.
Each scan records three kinds of observation: detected, unresolved and manual review. Issues are grouped by severity and by category, covering statutory content, policies, governance, finance, SEND, safeguarding, accessibility, data protection and more, each with the page, what was found and a recommended fix.
Design decisions
The scan never certifies a school's legal position; a person confirms applicability and evidence. It tells a requirement newly added to the list apart from one actually checked in the last scan. Issues can be marked in progress, resolved or reopened, and if a later edit reintroduces a problem, the next scan opens a new issue with new evidence while the old one stays in the trail. A batch of issues can go to the website provider as one neutral, branded email with each page and fix listed.
What you can see
A dated evidence record, a termly tracker across Autumn, Spring and Summer, and a working issue queue. The Policy Register sits beside it, and the Trust Website Audit applies the same approach to a trust's central site.
The result
The board can see what is published, what is unresolved and who is fixing it, with evidence it can inspect rather than a compliance percentage. The free public Website Check-up is a separate, lighter tool.