Ask a fleet manager what their vehicles cost to run and you will get an accurate answer, usually to the rupee or the dirham. Ask which drivers are responsible for the top and bottom of that range and the answer becomes an impression: this one is careful, that one is heavy on the brakes, this one we have had complaints about.
Impressions are not evidence, and they cannot be acted on. You cannot put an impression in a performance review, use it to set an insurance premium, or defend it when the driver disagrees. So the conversation does not happen, the behaviour continues, and the cost stays in the fuel bill where nobody attributes it to anyone.
Four behaviours, all measurable
Telematics hardware already reports what is needed. The question is whether anything computes with it.
- Idling — engine running, vehicle stationary, ignition on. Fuel burned to move nothing.
- Harsh acceleration — rapid throttle application, measured by accelerometer or by rate of speed change between position reports.
- Harsh braking — the same in reverse, and an event that should be reviewed in its driving context.
- Overspeeding — not the device's factory threshold, but speed against the limit configured for that vehicle class and route.
Each is a discrete, timestamped event attributable to one driver on one trip. Once they are counted per driver rather than per vehicle, a fleet stops having a fuel problem and starts having a small number of specific, addressable conversations.
The arithmetic, with your numbers
Idling is the easiest to quantify because the assumptions are few and you can check every one against your own records. The figures below are illustrative — substitute yours.
Idling cost, worked example
| Item | Value |
|---|---|
| Fleet size | 100 vehicles |
| Idling per vehicle per day | 45 minutes |
| Fuel burned while idling | 2 litres per hour |
| Operating days per year | 300 |
| Litres burned idling per year | 100 × 0.75 h × 2 L × 300 = 45,000 L |
Multiply the last line by your own fuel price. The point is not the number — it is that the number exists and is currently unattributed. A one-third reduction would save 15,000 litres in this example; it is an illustration, not a forecast or guaranteed result.
Run the same calculation with your real idling average. Most operators who measure it for the first time find the figure higher than they assumed, because idling is invisible: a stationary vehicle with the engine running looks identical to a parked one from the office.
Why scoring changes behaviour and reporting does not
A monthly report to management describes the past. It reaches the driver, if at all, weeks after the trips it describes, as a summary they cannot verify and did not see coming.
A score the driver can see on their own phone works differently. It is current, it is theirs, and it is comparable. The mechanism is not sophisticated — people modify behaviour that is visible and attributed far more readily than behaviour that is aggregated into a departmental total.
The design decision that matters
Show drivers their own score before you show managers the league table. A scoring system introduced as surveillance gets resisted and gamed. The same system introduced as feedback — here is your figure, here is the fleet average, here is what moved it — gets engaged with.
What honest measurement requires
- Thresholds set per vehicle class. A loaded truck and a light van do not brake alike, and scoring them against one threshold produces noise that drivers correctly dismiss.
- Events tied to a driver, not a vehicle. Without driver identification — an app login, a key fob, an assignment record — every event lands on whoever is usually in that vehicle, which is unfair and therefore unusable.
- A dispute route. Drivers will contest events, and sometimes they will be right: a genuine emergency stop scores identically to careless braking. A system with no way to annotate an event loses the drivers' trust the first time it is wrong.
- Consistency over time. A score that changes meaning when you adjust the thresholds is not a score. Fix the model, then leave it alone long enough for trends to mean something.
The second-order effects
Fuel is one measurable cost. Maintenance and incident costs should be evaluated separately using your own records.
Harsh braking and acceleration wear brakes, tyres, clutches and suspension at a rate that shows up in the maintenance budget rather than the fuel budget, which is why it is usually attributed to vehicle age instead of driving style. Collision frequency follows the same events closely enough that insurers in several markets will discuss premiums with an operator who can produce driver-level data — and being able to produce it after an incident, rather than arguing about it, has its own value.
None of this needs new vehicles, new depots or new headcount. It needs the data the devices already generate to be attributed to the people who generate it.
Key takeaways
- Idling, harsh braking, harsh acceleration and overspeeding are already reported by your hardware — the gap is attribution, not measurement.
- Calculate your own idling cost before buying anything; the number is usually higher than assumed because idling is invisible from the office.
- Driver-visible scoring changes behaviour; monthly management reports describe it.
- Set thresholds per vehicle class and give drivers a way to dispute events, or the scores will be dismissed as noise.
- Fuel justifies the project; maintenance and collision costs are usually the larger recovery.
Questions or a correction? Contact the 2M team. Examples explain a method; results depend on your operation and configuration.