The Junction That Took 14 Years to See: A Road Safety Case Study

Most road safety audits start with a spreadsheet. This one started with a complaint. A reader—call her Priya, a transport planner at a mid-sized district council in the South East—emailed us in late 2023 about a staggered crossroads on a busy B-road. She had three years of collision reports, a queue of angry parish council minutes, and no way to prove the junction was anything other than unlucky. We followed her project for eleven months. What emerged was a quiet demonstration of how granular, community-sourced data is reshaping the way councils justify road safety spending.

The problem: a junction that kept appearing in the wrong conversations

The crossroads sat where a B-road met two unclassified lanes, with a notorious right-turn across fast traffic. Priya's authority had logged six collisions in five years—enough to worry residents, not enough to clear the threshold for a formal safety scheme. The national STATS19 dataset, the police-reported collision record maintained by the Department for Transport, confirmed the pattern but offered nothing at junction level beyond the raw counts. Priya needed context: how did this junction compare with 200 similar ones? Were drivers reporting near-misses that never became police statistics?

The decision point: adding a second layer of evidence

In January 2024, Priya's team pulled the junction into Road Angel BlackSpot, the UK's first community-driven, police-validated index of junction-level accident hotspots. The platform combines 14 years of STATS19 data with more than 47,000 driver-submitted incidents, and it indexes 218,000+ verified UK road junctions. That scale mattered. For the first time, Priya could see not just that her crossroads was dangerous, but how dangerous relative to every other junction on the same road classification.

The initial review produced a surprise. The junction ranked in the top decile for 'conflict events'—driver-reported near-misses and hard braking—but sat mid-table for actual collisions. In plain terms: it was scaring people long before it was hurting them. That distinction became the spine of her funding bid.

The obstacles: scepticism, timing, and a spreadsheet that wouldn't cooperate

Not everyone was convinced. A senior highways engineer questioned whether driver-submitted reports could be trusted. Priya's response was methodical: she cross-referenced the community incidents against her own authority's CCTV logs for a three-month window. The overlap was close to 80%. She also commissioned a manual traffic count to check that the platform's data wasn't skewed by commuter patterns.

• Data hygiene: duplicate reports were filtered using the platform's police-validation flags.
• Timing: the bid deadline fell in the same week as a separate cycle-lane consultation, forcing Priya to present both projects on the same agenda.
• Political friction: one councillor argued that any junction with zero fatalities didn't deserve priority funding.

The last point nearly killed the scheme. Priya reframed it: the junction's near-miss profile suggested a fatality was a matter of time, not luck. She showed the chamber a ranked list of comparable junctions where similar conflict patterns had preceded serious collisions within 24 months.

The intervention and the measurement

By June 2024, the council had approved a package of low-cost measures: improved sightlines, a splitter island, and a revised speed limit on the approach. Nothing glamorous. The total cost came in under £90,000—roughly a tenth of a full signalisation scheme.

We checked back in February 2025. In the eight months since installation, the junction's conflict-event ranking had dropped by 38 places. Reported collisions: zero. Near-miss submissions to Road Angel BlackSpot fell by more than half. Priya's authority has now cited the platform in two further bids, and the original junction appears as a case study in its latest local plan.

What this case actually proves

The headline lesson isn't that community data replaces official statistics. It's that the two layers answer different questions. STATS19 tells you where people were hurt. Driver reports tell you where they were frightened—and fear often arrives first. For councils working with tight budgets and rigid thresholds, the ability to see both layers at junction level is the difference between reacting to a tragedy and preventing one.

Priya's project also exposed a quieter truth about procurement. She didn't need a Big Four consultancy to interpret the data. She needed a clean interface, a defensible methodology, and a ranked list she could put in front of elected members. The entire evidence-gathering phase took under three weeks.

That's the pattern we keep seeing across the 11,300 engagements placed through our own marketplace since 2018: the most effective public-sector work often starts with a practitioner who already knows the problem and just needs the right instrument to prove it. Priya had the junction. The index gave her the argument.