How a Score Is Built

The number at the top of a driver's report is an arithmetic construction, and the construction is rarely explained to the people it ranks.

The three parts

Event counts, by type: braking, acceleration, cornering, speeding, sometimes idling.

Weights, deciding how much a cornering event counts against a braking one.

Exposure, usually distance or time, which turns counts into a rate.

Score equals weighted events divided by exposure, scaled. That is the whole model in most products.

What the weights encode

A value judgement about which behaviours matter, made by the vendor.

Frequently without evidence linking the weights to collision risk in your kind of operation.

Ask what the weights are and where they came from. A vendor who cannot say has built an opinion and presented it as a measurement.

And ask whether you can change them, because a fleet with a specific risk profile may reasonably weight differently.

The exposure problem

Distance is the usual denominator and it is the wrong one for urban work.

A driver doing forty drops in twenty miles has far more opportunities to brake than one doing two hundred motorway miles.

Per-mile scoring therefore penalises urban rounds systematically, by construction.

Time is better for stop-start work; events per hour of driving, or per stop, better still where the data allows.

Check which your product uses and whether it can be changed.

Route type dominates

In most fleets, the largest variable in the score is the round rather than the person.

Which is testable: swap two drivers' rounds for a fortnight and watch the scores swap with them.

Few fleets run this test and it settles the argument faster than anything else.

What a score is useful for

A driver compared against their own history, which removes route and vehicle effects.

A round compared across time, which identifies a route that generates events.

Identifying an outlier for a conversation, with the context checked first.

Not: a league table, a bonus calculation, or a disciplinary input on its own.

The question to ask a vendor

"Show me the formula."

Weights, exposure basis, normalisation, and whether any of it adapts over time.

If the answer is that the model is proprietary, the score cannot be defended to the person it describes — and that is the situation that ends in a grievance you cannot answer.

Run the round-swap test

The experiment that settles the argument about routes.

Swap two drivers' rounds for a fortnight.

If the scores swap with the rounds, the score is measuring the work.

Few fleets run this, and it resolves in two weeks a dispute that otherwise runs for years and ends in a grievance.

Follow the difficult record

Use workforce optimisation software for driver teams to frame one representative case. The useful evidence is the record created when a value is challenged, corrected, approved and exported.

Independent reference

For a thematic point of reference, see the Insurance Institute for Highway Safety. Consult the source directly because technical, operational and legal details change.