Fleet safety programmes fail in a recognisable way: a policy document, an annual toolbox talk, a telematics dashboard nobody acts on, and a collision rate that does not move.
Programmes that work share a structure.
The components
1. Leadership commitment that is visible. Not a statement in a handbook. Senior managers who ask about safety in operational meetings, who accept a service failure caused by a driver refusing an unsafe vehicle, and who fund the programme.
2. Risk assessment specific to your operation. Urban multi-drop, motorway trunking, rural service and construction site delivery have different risk profiles. A generic assessment produces generic controls that address nobody's actual risks.
3. Written policies people can follow. Speed, mobile phone use, fatigue, alcohol and drugs, reversing, loading, personal use, and what to do after an incident. Short, specific and enforced consistently.
4. Driver selection and assessment. Licence and record checks at hire and periodically, competence assessment for the vehicle class, and induction that covers your actual operation.
5. Training that is targeted. Generic annual training changes little. Training triggered by risk — a driver with a pattern of reversing incidents, a depot with a spike in low-speed collisions — changes behaviour.
6. Vehicle safety specification. Reversing cameras and sensors, blind spot detection, automatic emergency braking, direct vision. Specification decisions at purchase remove risk for the vehicle's whole life, which is a better investment than most training.
7. Monitoring. Telematics behaviour data, video events, incident reports, near-miss reports, inspection findings, and public complaints.
8. Investigation. Every incident investigated for cause, not blame, with corrective actions tracked.
9. Review. Quarterly review of the whole programme against outcome data, with changes made.
In most delivery fleets, low-speed manoeuvring incidents outnumber every other category — usually a large majority of all collisions.
Measuring what matters
| Metric | Notes |
|---|---|
| Collisions per million kilometres | The core rate; normalise or it is meaningless |
| At-fault collision rate | Separates what you can control |
| Collision cost per vehicle per year | Includes repair, downtime, third-party, excess |
| Near-miss reports | A rising rate is usually good — it means people are reporting |
| Incidents by type and location | Where to focus |
| Behaviour event rate, normalised | Leading indicator |
| Time from incident to investigation closure | Programme discipline |
Near-miss reporting deserves emphasis. A rise indicates trust and engagement, not deterioration. Programmes that punish near-miss reports get zero of them and lose their best early warning.
Investigation that produces change
The point is to find the system failure, not the person to blame.
For each incident ask:
- What happened, factually, from the evidence?
- What were the immediate causes?
- What conditions made those causes possible — schedule pressure, vehicle condition, site design, training gap, unclear procedure?
- What would prevent recurrence, and is it within our control?
- Who owns the action, and by when?
A blame-focused investigation produces a disciplinary record and no learning. Drivers stop reporting, near misses disappear, and the same incident happens again in six months at a different depot.
Post-incident driver support
Drivers involved in serious incidents, particularly those involving injury, need support. Beyond the ethical obligation, an unsupported driver is a driver who is more likely to leave and less likely to report the next incident. Have a defined process: immediate welfare, time off if needed, access to support services, and a return-to-work assessment.
Contractors and agency drivers
Programmes frequently exclude the drivers with the highest risk profile. Agency and subcontracted drivers are typically less familiar with your vehicles, your sites and your procedures.
Include them: induction before first use, the same monitoring, the same incident investigation, and contractual requirements on the agency for selection and training standards.
Frequently asked questions
What collision rate should we target?
Rates vary enormously by operation type, so benchmarking against other fleets is of limited use. Set targets against your own trend, segmented by incident type, and focus on the categories that dominate your cost — usually low-speed manoeuvring.
Does driver training reduce collisions?
Targeted training addressing specific identified risks does. Generic annual classroom training has repeatedly shown weak effects. Use monitoring data to identify who needs what, and deliver it close to the behaviour it addresses.
Should we install cameras?
Video telematics with a structured coaching programme reduces incidents in most fleets that implement it properly, and it protects drivers in disputed collisions. It also carries privacy obligations and consultation requirements — see video telematics.
How do we build a reporting culture?
Respond constructively to every report, never punish someone for reporting, close the loop visibly, and demonstrate that reports produce changes. Culture follows what happens after the first difficult report, not what the policy says.
What is the cheapest effective safety intervention?
Usually vehicle specification at purchase — reversing cameras, sensors, and blind spot systems — because it removes risk permanently without depending on human behaviour under pressure. Site and route design that eliminates unnecessary reversing is a close second.
