Reviewing Footage

A camera system generates clips. Someone watches them, decides what they mean, and acts. That someone is the programme.

The volume problem

A fleet of a hundred vehicles with event triggers produces more clips than anyone expects.

Most are false: a pothole, a kerb, a defensive manoeuvre that was correct.

Which means the reviewer's first job is dismissal, and the rate of dismissal is the number to watch.

A programme where every clip becomes a conversation will collapse within a quarter.

Who reviews

Someone trained to distinguish a good defensive manoeuvre from a lapse, which is a skill and is not obvious.

Not the driver's direct supervisor, ideally, because that conflates safety review with line management.

And not an outsourced reviewer with no knowledge of the route, which produces findings that a local manager then has to unpick.

Where the fleet is small, it is the fleet manager, and the training matters more.

Reading a clip fairly

Watch the whole buffer, not the event moment.

The cause is frequently before the trigger: a vehicle cutting in, a pedestrian, a signal change.

A hard brake that avoided a collision is good driving that scored badly, and treating it otherwise teaches drivers to brake later.

Check the vehicle and the load, because both change what a given deceleration means.

What to record

Reviewed, and the outcome: dismissed, noted, coached, escalated.

With a reason for anything above dismissed.

Because a driver asking why they were coached is entitled to an answer that is not "the system flagged it".

Telling the driver

That their footage was reviewed, and what the conclusion was — including when it was dismissed.

This is the practice that distinguishes a safety programme from surveillance, and it is the one most often skipped because it takes time.

A driver who only hears from the system when something is wrong will experience it as punitive, accurately.

The measures

Clips generated per vehicle per week.

Dismissal rate, which if very high means the triggers need tuning and if very low means the reviewer is rubber-stamping.

Time from event to review, because a clip reviewed three weeks later is useless for coaching.

And review outcomes distributed by driver, watching for concentration that reflects route rather than person.

Watch the whole buffer

The discipline that makes review fair.

The cause is usually before the trigger: a vehicle cutting in, a signal, a pedestrian.

A reviewer who watches only the event moment sees a hard brake and not the reason for it.

And a hard brake that prevented a collision is good driving that scored badly, which is the judgement the whole review exists to make.

A practical configuration prompt

During configuration, use view the product overview to prompt questions about purpose, ownership and output. Treat the page as a starting point and document each assumption.

Independent reference

For a thematic point of reference, see the Information Commissioner's Office. It provides useful context beyond supplier documentation and should be checked in its current form.