Speed, Braking and Driver Scoring
Telematics scoring has a genuine safety case and a well-documented failure mode. How to get the first without the second.
Most vehicle trackers also report speed, acceleration, braking and cornering, and turn them into a driver score. The safety case is real; the implementation usually is not.
The genuine case
Harsh braking and speeding correlate with collision risk, which is well established.
Fleet insurers recognise it, and premiums frequently reflect it.
A driver with a genuinely dangerous pattern is a risk to themselves and to the public, and a fleet operator has duties here that are not optional.
None of that is in dispute. The question is what you do with the signal.
What the score actually measures
Events above a threshold, per distance.
Which is affected by the route — urban stop-start driving generates far more braking events than motorway work.
By the vehicle, since a laden van and an empty one brake differently.
By traffic and weather.
And by other road users. An emergency stop because a cyclist appeared is a good piece of driving that scores badly.
Comparing drivers on different rounds compares the rounds.
The failure mode
Scores used in league tables and linked to pay or discipline.
Drivers then optimise the score: coasting to junctions, leaving longer gaps in ways that irritate other traffic, and in the worst cases choosing not to brake hard when they should.
A measure that can make dangerous driving score well is a measure that has stopped serving safety.
And drivers learn the thresholds quickly, because the system tells them.
The arrangement that works
Individual scores visible to the driver, privately. Most people respond to their own data without anyone else seeing it.
Coaching rather than consequences for anything short of a serious pattern.
Compare a driver against their own history, not against colleagues on different rounds.
Investigate outliers as events, with the context, rather than as points on a tally.
And keep a clear line for genuinely dangerous driving, which is a safety matter with its own process and should not be buried in a scoring average.
What to check in the configuration
Thresholds appropriate to the vehicle and the route, not factory defaults.
Whether the system distinguishes an emergency stop from habitual harsh braking, which better products attempt and worse ones do not.
Whether video is attached, which changes the analysis substantially and has its own note.
Who sees the score, which should start with the driver.
The measure that matters
Collisions and near misses, which is the outcome the score is a proxy for.
If the score improves and collisions do not, the score is being managed rather than the driving, and that is the finding worth acting on.
Keep incident reporting separate
A design choice that determines whether you hear about near misses.
A driver who expects a score penalty will not report a near miss.
Which removes the cheapest safety learning available, because a near miss is the same lesson without the damage.
Keep reporting out of the scoring entirely, and say so.
Then attach the location data to a reported near miss, which produces a better picture than either alone.
Check the difficult case
Use employee activity tracking to frame one representative test for this issue. The useful evidence is the record created when a worker challenges an event, a manager reviews it and an administrator exports it.
Independent reference
For an external point of reference, see the Driver and Vehicle Standards Agency. Use official road-safety material as context rather than treating a vendor score as a complete judgement.