Methodology
The number is arithmetic. The sentence is the record.
A 0–100 on a title. Base 70, then signed deltas that sum back to the score. Every rate carries its denominator. Under five decided appeals the rate is withheld and the page says too few.
13
scored constraint layers from planning.data.gov.uk, the Environment Agency, Natural England and Historic England
12
refusal grounds, read from the Inspectorate’s decision letters and counted per council with the sentence behind each
500 m
of decided applications around every point, refusals first, each linked to the council’s own record
0.71 AUC
with its interval, 0.63–0.79, published per authority and re-run each quarter
From a point to a verdict
01
The publisher answers first
A per-site check hits the authoritative register by id. Our copy is the cache and the fallback. A filled dot means live; a hollow one means copy and its age.
02
Constraints become signed deltas
Base 70. Each scored layer moves the number, and the moves sum back to the score on the page. A missing layer is not proof of no constraint where the publisher holds nothing for that authority.
03
The letters explain the number
Twelve refusal grounds, counted from the inspectors’ own sentences, mix-adjusted for the council, withheld under five. The page describes the public record. It does not predict permission.
What is the catch at £0?
Almost everything Plot reads is open data: planning.data.gov.uk, HM Land Registry’s INSPIRE polygons, the Planning Inspectorate’s decision letters. What costs money to make is the classification of those letters, the join to your parcel and the calibration behind the score, and that is what the paid plans carry across more than one authority. Ownership names, comparables and landowner letters are not here; the sourcing platforms do those well.
Not planning advice
Named layers live on Data sources.