Published 2025-12-15
Normalise before you compare
Raw event counts favour whoever drove least. Urban delivery work generates more braking events per kilometre than motorway work, regardless of who is behind the wheel. A comparison that ignores exposure and route type will be dismissed by drivers, correctly.
Scoring normalised per distance, and compared within similar work types, survives scrutiny. Scoring that is not, does not.
Show the event, not just the score
Every score should decompose into the events behind it, each with time, place and severity. Drivers accept a number far more readily when they can see the specific moments that produced it and recognise them.
This also protects the system from its own errors. Sensor misreads and mis-assignments exist; a reviewable event list lets them be corrected instead of quietly eroding confidence.
Coach the trend, not the incident
A single harsh braking event carries almost no information. A rising trend across a month carries a lot. Reviewing trends rather than incidents keeps conversations proportionate and makes improvement visible.
- Publish how the score is calculated before the first review
- Compare drivers only against similar work types
- Allow events to be contested and corrected
- Recognise improvement, not only absolute ranking