Road Angel BlackSpot
Est. Manchester, 2011
Verified dataset — Tuesday 14 January
Methodology White Paper · Edition 2025.1
The BlackSpot Score is a 0–100 index of junction-level road danger, built from 14 years of STATS19 police collision data and cross-checked against 47,800+ community reports each year. This document is the authoritative methodology cited by 31 local councils and 14 peer-reviewed papers — the reference researchers and council officers bookmark when they need to know how a number on our map was reached.
Abstract
The BlackSpot Score assigns every mapped junction in Great Britain a single integer from 0 (lowest recorded danger) to 100 (highest severity, weighted by casualty outcome). It is maintained by Road Angel BlackSpot Ltd, a Manchester-based publisher founded in 2011 by former Metropolitan Police traffic analyst David Hargreaves. The score fuses five weighted inputs: police-reported collision severity from STATS19, normalised incident volume, community confirmation counts from driver submissions, casualty density per traversed vehicle, and an infrastructure risk score audited by the company's in-house engineering team.
The index is recomputed every Tuesday at 04:00 GMT against the most recent police upload window. A junction's score is published only after it has cleared a 48-hour STATS19 cross-reference and has gathered at least three independent community confirmations or one corroborated police report. The methodology described here is the version operative from 14 January 2025; earlier schema notes are archived at /methodology/archive.
§02 · The Formula
A BlackSpot Score is the normalised sum of five evidence streams. Each input is rescaled to a 0–100 sub-score before weighting, so a sparsely-reported rural junction and a heavily-trafficked London roundabout are compared on the same axis. The final integer is published to two significant figures.
Weighted casualty outcome from STATS19 collision records within a 50-metre radius of the junction centroid over the trailing 36 months. Fatal and serious-injury collisions carry 10× the weight of slight-injury collisions; damage-only reports are excluded.
Total STATS19 collisions within the same radius and window, normalised against the national median for the junction's road class (motorway, A-road, B-road). A junction above its class median is penalised in proportion.
The count of independent driver-submitted incident reports within the trailing 12 months, discounted by the platform's confirmation rate when cross-checked against police logs (currently 94%). Reports older than 12 months decay exponentially.
Casualties per million traversing vehicles, estimated from DfT traffic counts and the junction's recorded collision history. This term corrects for the raw volume bias that would otherwise favour the busiest junctions irrespective of risk.
An audited score from the company's engineering panel (six ex-police collision investigators and two former Highways England engineers) covering sightline, surface, signage, and layout factors. The audit is re-run every 24 months or after any reported improvement.
A score below 30 is not published. A score at or above 70 carries the active "BlackSpot" warning visible on the public map. Scores are recalculated only when a new input event occurs; static junctions retain their last weekly value.
§03 · Data Sources
Every published BlackSpot Score rests on at least one primary feed and is independently corroborable against the others. The feeds below are listed in descending order of statutory authority; the platform treats each as an auditable input, never a substitute for the others.
The UK national reporting system for personal-injury road collisions, completed by police officers at the scene. Road Angel BlackSpot ingests the full STATS19 release within 14 days of publication and retains a 14-year rolling window. This is the only feed that can, on its own, raise a junction above the publication threshold.
Driver-submitted reports via the website and the 312,000-download mobile companion. Each report carries a hashed device ID, a coarse location pin, a free-text description and an optional photograph. Reports are aged out at 12 months unless re-confirmed; current cross-check confirmation rate against police logs is 94%.
Injury logs supplied directly by partner council road-safety teams under formal data-sharing agreements. Used as a corroborating feed, particularly for pedestrian and cyclist casualties that are under-represented in STATS19. Currently feeds from 31 of the 36 English metropolitan and unitary authorities.
Manual site audits of sightlines, surface condition, signage and layout, conducted by the in-house engineering team. Audits are also triggered by community reports flagging a recent change (e.g. a removed mirror, a resurfacing scheme). Audits feed only the I term of the formula and cannot raise a junction above score 50 on their own.
§04 · Worked Example
The junction of the A232 Addington Road and Gravel Hill, Croydon, was the catalyst for the platform's founding in 2011 after David Hargreaves's sister was injured there. The trace below is reconstructed from the platform's public audit log and illustrates how a junction moves through the pipeline.
Tuesday 07:14 GMT · Driver report received
A commuter driving southbound on Gravel Hill at 06:48 submits a report via the iOS app describing a near-miss at the Addington Road junction during the morning rush. The report carries the device's hashed ID, a 12-metre-precise pin dropped on the carriageway, and a 38-word free-text description. The platform flags it for the Tuesday 04:00 recompute window.
Tuesday 04:00 GMT · Weekly recompute
The platform pulls the latest STATS19 release, layers in the new submission, and computes a candidate score. At this stage the junction sits at B = 41 — driven mostly by the existing P and V terms, which already carry four police-reported collisions in the trailing 36 months. The C term lifts the score modestly; K and I are unchanged from the previous week.
Wednesday 14:22 GMT · 48-hour STATS19 cross-reference
The platform's matching engine compares the new submission against STATS19 records for the same 50-metre radius in the trailing 90 days. A police collision report from 09 January — a slight-injury two-vehicle shunt — is matched on coordinates and time-of-day to within the platform's tolerance. Confirmation rate against police logs for this junction moves to 96%.
Tuesday 14 January 2025 · 04:00 GMT · Publication
The recompute confirms the junction at B = 47 — well below the 70-point BlackSpot threshold, but firmly above the 30-point publication floor. The pin turns red on the public map, the postcode-level JSON feed refreshes within the hour, and the 140,000+ dashcam-equipped subscribers receiving the free API are pushed the updated tile. The score will next recompute on 21 January, or sooner if a fresh event is logged.
§05 · Institutional Adoption
The BlackSpot Score is referenced in published research, council policy documents, and a formal data-sharing partnership with National Highways (formerly Highways England). The figures below are the live measures of that adoption as of January 2025.
The BlackSpot Score is referenced in the published road-safety plans of 31 local councils across England and Wales, including adopted local transport plans and casualty-reduction strategies. Councils receive de-identified hotspot tiles on request.
Cited in 14 peer-reviewed road-safety papers since 2018, including a 2021 Transport Research Board paper on community hazard reporting and a 2023 Journal of Transport & Health study on junction-level severity modelling.
Formal data-sharing partnership signed in 2019 with Highways England (now National Highways) under which de-identified hotspot data feeds into the M-roads safety review. Partnership does not constitute endorsement of any score.
Across the platform's lifetime — 1.2 million community-submitted hazard reports — 94% of corroboration attempts against police logs have matched within tolerance. This figure is the platform's central reliability claim and is recomputed annually.
The methodology described above is the operative schema from 14 January 2025. Researchers and council officers who wish to cite this document should use the DOI registered with the British Library's ETD archive under the reference RA–BS/METH/2025-01.
Report a BlackSpot Near You