August 2026 Edition

Methodology, denominators, and what we don't yet cover

• Window. Headline figures are trailing 12 months ending 31 Aug 2026 (the most recently completed calendar month). Year-on-year (YoY) deltas compare against the immediately preceding T12M window of the same length. The "last 90 days" callout is the same data restricted to that shorter window — included for currency, not as a primary figure. T12M is the headline because it smooths planning's strong seasonality (Christmas dip, summer surge) and gives borough-level samples large enough to rank with confidence.
• Headline metric — applications received. Every planning headline is built on applications received in the window. The "received" date is the LPA's official submission date (date_received) when the council portal exposes it, falling back to the LPA's validation date (validated_date) when only the list-page date was captured — typically 0–7 days after receipt and published by the same council on the same portal. Both are LPA-attributed dates, not our observation date. Decisions are a sub-breakdown — the closed portion of the pipeline. We lead with received because it's the broadest defensible measure a critic with the same portals open could verify.
• Headline sanity-check gate. Before this edition was published, every headline figure was checked against the raw database via a deliberately different query path, against MHCLG Live Table P120 (the national authoritative per-LPA application volume figures), and against the prior edition's number. Any borough's count below 50% or above 200% of P120 expected, or any London-total swing exceeding ±40% month-over-month, blocks publish. The check failure email is treated as a P0 incident.
• Approval rate denominator. Approval rate = approved / (approved + refused). Withdrawn applications are excluded from the denominator because they're not council decisions — the applicant pulled them. Some councils encourage withdrawal to avoid recording a refusal; we surface the withdrawn count separately so the reader can see if the pattern is unusual.
• Borough leaderboard guardrails. A borough is only ranked on approval rate if it meets both floors: at least 50 decided in the window (sample size) AND at least 50% of its received applications classified (representativeness). The second floor is critical — a borough with 50 decided out of 1,200 received is a 4% sample, and ranking off that risks publishing a number a critic could disprove with the council portal open. Each ranked row shows both numbers; excluded boroughs are counted separately so we don't hide exclusions.
• Borough coverage and publishable set. Astor automates daily collection from 23 of London's 33 council planning portals. A borough enters the published headline only when our T12M coverage reaches at least 50% of MHCLG Live Table P120's expected receipts for that borough. This edition covers 30 borough(s) at that confidence level.
  • Barking and Dagenham — 1243 received, 161.3% of P120 expected
  • Barnet — 5217 received, 189.3% of P120 expected
  • Bexley — 2287 received, 163.9% of P120 expected
  • Brent — 3153 received, 143.9% of P120 expected
  • City of London — 1378 received, 255.8% of P120 expected
  • Croydon — 3248 received, 187.0% of P120 expected
  • Ealing — 4041 received, 194.3% of P120 expected
  • Enfield — 4370 received, 218.0% of P120 expected
  • Greenwich — 3126 received, 208.0% of P120 expected
  • Hackney — 2457 received, 176.0% of P120 expected
  • Hammersmith and Fulham — 2400 received, 162.3% of P120 expected
  • Haringey — 2190 received, 151.9% of P120 expected
  • Harrow — 1818 received, 123.3% of P120 expected
  • Havering — 2402 received, 154.3% of P120 expected
  • Hillingdon — 2576 received, 142.6% of P120 expected
  • Hounslow — 2554 received, 179.2% of P120 expected
  • Islington — 3091 received, 197.5% of P120 expected
  • Kingston upon Thames — 2557 received, 220.6% of P120 expected
  • Lambeth — 3263 received, 169.4% of P120 expected
  • Lewisham — 2371 received, 175.1% of P120 expected
  • Merton — 1574 received, 129.5% of P120 expected
  • Newham — 2555 received, 218.2% of P120 expected
  • Redbridge — 3509 received, 195.8% of P120 expected
  • Richmond upon Thames — 4189 received, 213.3% of P120 expected
  • Southwark — 3645 received, 210.7% of P120 expected
  • Sutton — 1789 received, 186.9% of P120 expected
  • Tower Hamlets — 1809 received, 172.9% of P120 expected
  • Waltham Forest — 2169 received, 153.1% of P120 expected
  • Wandsworth — 4206 received, 177.6% of P120 expected
  • Westminster — 9156 received, 186.5% of P120 expected
  The following automated boroughs are excluded from this edition's headlines because our list coverage is below the 50% threshold — publishing their T12M counts would report a function of our coverage gap, not real activity:
  • Bromley — 0 received vs 2347 expected (0.0%)
  • Camden — 749 received vs 2317 expected (32.3%)
  • Kensington and Chelsea — 1126 received vs 2649 expected (42.5%)
  The following are not yet automated and may be entirely absent from this report's numbers:
  • Barking and Dagenham — BeFirst portal blocks automated requests; under review
  • Bromley — Bespoke council portal; integration in progress
  • Hackney — Bespoke council portal; integration in progress
  • Haringey — Salesforce-backed portal; bespoke integration needed
  • Harrow — Integration not yet available
  • Hounslow — NECS portal; bespoke integration needed
  • Kensington and Chelsea — Bespoke council portal; integration in progress
  • Merton — Regulatory Hub portal; bespoke integration needed
  • Redbridge — Idox SwiftLG variant; integration in progress
  • Waltham Forest — BeFirst portal blocks automated requests; under review
• Classification coverage. 89.2% of applications received in this window have been status-classified by our pipeline. The remainder are awaiting detail collection and counted as "unknown". When we cite approval rates, the denominator is the classified subset — so for boroughs with low classification, our ranking floor (above) keeps them out of leaderboards.
• Sales. HMLR Price Paid, PPD category A only (standard residential transactions; excludes auction sales, repossessions, and buy-to-let transfers). HMLR publishes with a ~2-month lag, so any month after Jul 2026 is preliminary in the snapshot. The median-resale headline excludes new builds because their pricing is developer-set, not market-clearing; the all-transactions median is computed too but is the volatile one.
• Article 4. "Designated in T12M" uses the legal start_date of each direction (not the date we first stored the row). Active count = directions whose end_date is null in our records. The Open Government Licence v3 feed at planning.data.gov.uk is the primary source; we hand-curate a small seed for borough-wide HMO directions known to be missing from the national feed (1 of this window's 18 additions came from that seed). A zero count for a borough does not mean that borough has no Article 4 — verify with the council.
• HMO classification. Per-application HMO signal set by Astor's triage (Claude Haiku 4.5) reading the proposal text — see the Planning Checker for the per-application view. The HMO-flagged decisions count compounds three pipelines (list collection × detail collection × triage), so its absolute number is conservative; treat as a floor, not a measurement.
• AI commentary. Generated by Claude Haiku 4.5 with structured tool-use; every number it cites must appear on this page. Commentary is advisory and editorial, not regulated financial or planning advice. Astor is not FCA-authorised.
• Scope & sourcing. Planning figures cover the London boroughs we track end to end; residential sales cover all 33 London boroughs. Every figure is traced to a public source — HMLR Price Paid, council planning portals, and planning.data.gov.uk. A zero or an absence means we hold no record of it, not that none exists — verify with the relevant council before relying on any specific finding for a decision.