Where to Start

Most programmes start with a product demonstration. The order that works starts with your own incident history.

Weeks one to four

Collect three years of collisions and claims.

Classify by type: low-speed manoeuvring, junction, rear-end, vulnerable road user, motorway.

By tenure: how many involved drivers in their first six months.

By time of day and by round.

No hardware required, and it usually reframes the project — most fleets discover their problem is reversing damage and new starters rather than harsh driving.

Weeks five to eight

Check what the vehicles already have. Emergency braking, lane keeping, sensors, direct vision.

Check the schedule: drive a difficult round yourself and time it honestly.

Check the rota for fatigue patterns: consecutive nights, turnarounds, start times.

Ask drivers where the dangerous places are, which takes an afternoon and produces a map.

Weeks nine to twelve

If the incident profile points at behaviour, trial a system.

A month of silent logging, thresholds from your own distribution.

Review twenty clips and count the false triggers.

Then decide, with the false positive rate and the workload known rather than assumed.

What to avoid

Buying before the incident analysis, which is how a fleet with a reversing problem ends up with a scoring package.

Announcing before thresholds are set.

Starting with cab-facing cameras, which is the hardest sell and rarely the first thing the analysis supports.

And treating the demonstration's event examples as representative, because they were chosen.

What a quarter should produce

An incident profile by type, tenure, time and round.

A list of structural findings: a junction, a round, a vehicle type, a rota pattern.

A vehicle specification recommendation for the next purchase cycle.

And a decision on monitoring made on evidence, which may be that forward-facing cameras and a schedule change address most of what the analysis found.

Start with the incident history

Not with a demonstration.

Three years of collisions and claims, classified by type, tenure, time and round.

No hardware required.

It frequently reframes the whole project: most fleets discover their problem is low-speed manoeuvring damage and new starters rather than harsh driving — and those need different answers.

Reproduce the workflow

For another way to make this requirement testable, consult view the form-based example. Reproduce the case with real roles, failures and recovery steps.

Independent reference

For a thematic point of reference, see the Royal Society for the Prevention of Accidents. This established source is a useful independent reference for the issue discussed here.