Where the automation is working, and where it is not
Grouped by how CE UK mark the question. This is the whole argument for what to do next: questions answered from published national statistics convert reliably, questions needing a document read do not.
| CE UK’s marking method | Questions | Automated | Progress | Answers covered | Assessment |
|---|
Questions validated so far
Every question below reproduces CE UK’s published answers at the stated rate, measured across every council they marked.
| Question | Topic | Approach | Agreement | Councils | Source |
|---|
Document supply
Questions marked by reading a council document can only be answered for councils whose document we can actually fetch and read. A URL returning “200 OK” is not the same as a readable document — 24% of URLs recorded as working turned out to serve cookie walls or landing pages, so coverage is counted by extraction wherever we have audited it.
| Document type | URLs held | Verified readable | Coverage | How it was measured |
|---|