Methodhow this is collected, and what it cannot tell you
The premise
Public portals publish a reading, not a record. The number you see is the number as of now; yesterday's is gone. That is not a conspiracy — it is how the systems are built — but it means the questions that matter most cannot be asked from the portal itself. Is this getting better or worse? Did that figure change quietly? Is one district being served differently from the one next to it?
So we take a copy every day and keep it.
How a figure gets here
The page is fetched and stored exactly as served. Nothing is parsed at this stage, so a bug in our reading can never cost a day of history.
Each row is matched to a district through the government's own Local Government Directory codes, so sources can be compared with one another.
A separate, re-runnable pass reads numbers out of the stored copies. Fix the reader, re-run it over the whole archive.
One question per page, answered in a sentence, with the capture time and the stored file named underneath.
Rules we hold ourselves to
Never "as of today". Every figure carries the moment it was captured. You will not find those words anywhere on this site, because they would not be true.
Where the source gave nothing, we do not write zero. You will see —
instead. Empty and zero are different claims.
Derived numbers are labelled derived. When we add districts up ourselves, we say so, and we say when the result differs from the portal's own total.
We publish what we got wrong. Corrections change the page and are noted on it, rather than being quietly replaced.
What we do not collect
District-level aggregates only. No personal records, no case-level rows, no names — none are collected, and none are published.