Methodology

Rasmere reports describe what has historically happened on comparable sites in the same Local Planning Authority. They are evidence summaries, not forecasts, and they are not planning advice. Every figure on the page can be traced back to the specific decisions that produced it.

Where the underlying evidence is thin, we say so on the page rather than smoothing the number. A 22% approval rate from 9 decisions is shown with the sample size visible.

The address you enter is normalised through Google Places (for coordinates) and the Office for National Statistics postcode service (for the admin_district attached to that postcode). That district is matched to the corresponding Local Planning Authority in our LPA registry.

If a postcode does not resolve to a known LPA, the report falls back to nearest-neighbour matching by coordinate. No five-borough hardcoded shortcuts exist; coverage is national.

Comparable cases are pulled from our planning case index within a fixed radius (default 500m) of the resolved coordinate. We do not silently expand the radius to inflate the count — if there is nothing within 500m, the report shows the empty state and explains why.

When an application-type filter wipes the result set, the system retries once at the same radius without the type filter, and clearly labels the broader sample.

Approval probability is the proportion of decided comparable cases (decisions containing "approved", "granted" or "permit") over all decided comparable cases (which also includes "refused" and "rejected"). Withdrawn and pending applications are excluded from the denominator.

We display the sample size next to every probability. A probability based on fewer than 20 decided cases is flagged as low-confidence.

Officer approval rates are computed on the case officers named in the public decision notice for each comparable. An officer with fewer than ten decided cases in the area is omitted from the league table — small samples are noise.

Officer-level figures are descriptive, not predictive. They reflect the historical mix of applications routed to that officer, not a fixed personal disposition.

Refused decisions are tagged into a fixed taxonomy (e.g. massing, amenity, design, policy conflict, highways) by a server-side categoriser that reads the officer reason for refusal. Tags are stored alongside the decision and surfaced as the refusal pattern bars in the report.

The categoriser is deterministic; the same input always produces the same tag set. Tags are reviewed against new policy language and updated as the underlying records evolve.

The national feed is drained on a one-minute scheduled job that pulls newly published decisions across England via PlanIt and a per-LPA incremental scraper queue. Cases reach the index within a small number of minutes of publication on most authorities.

Backfill is run separately. The current historical horizon extends back to 2018 and continues to expand as the queue drains.

Rasmere does not give planning advice, does not submit applications on your behalf, and does not predict the decision on your specific scheme. It will not invent comparable cases, fabricate officer names, or rewrite refusal reasons. If a panel has no data, it shows an empty state — not a placeholder figure.