9 mins read
A constant stream of behavioural data: speed, braking, acceleration, idling, and fuel burn are generated daily for fleet vehicles, logged in real time by telematics units, onboard diagnostics and GPS tracking. The data may be abundant, but is it telling fleet managers what they think it’s telling them?
New research suggests that a low driver score does not always indicate a bad driver. Instead, it may reflect the route, traffic, payload, vehicle, delivery schedule or customer-site conditions as much as the person behind the wheel. There are three questions telematics should answer: 1) where is cost or risk occurring? 2) what operating conditions sit behind it? and 3) which intervention is most likely to reduce it.
Where behaviour really moves the needle
Harsh acceleration, sharp braking, excess idling and poor anticipation all waste energy, and whether driving an ICE (Internal Combustion Engine) vehicle or an EV (Electric Vehicle), the result is the same. More fuel is consumed or less range is achieved from every charge. On the face of it, telematics data for real-world freight operations consistently show that driving habits affect fuel consumption and emissions. The picture is however, more nuanced.
Driver scores are shaped by operating conditions as much as by behaviour. An understanding of the impact of payload, route, congestion and scheduling allows a fleet to measure more than driver performance. A driver in a lightly loaded van on the motorway will never produce comparable numbers to a driver with a fully laden van whose regular route is in city traffic. Without monitoring similar vehicles, similar loads, similar routes and similar schedules, driver data may be telling fleets a lot less than they think.
When the route is the issue
Research published in Scientific Reports in 2024 analysed information gathered from connected vehicles over a two year period from 2016 and 2018. The data was based on 354,000 road segments and over 17,700km of road network across the West Midlands, UK and tracked idling, cruising, acceleration and deceleration. According to the analysis, idling and other driving measures varied by time and location, with idling increasing throughout the afternoon as congestion worsened.
Clearly, if several drivers’ scores are reducing on the same roads, and at the same time, the most likely explanation lies in the route, delivery window or customer-site conditions and not simply individual behaviour.
Behaviour shows up again in the maintenance bill
A DriveTech whitepaper, The Impact of Driver Behaviour on Vehicle Running Costs, found that persistent harsh braking or cornering was associated with increases of up to 73% in annual tyre costs in its analysis of 2,167 AA vans. And it’s not just braking that has a tangible cost. driving style directly impacts tyre life, brake wear, servicing frequency and increased insurance costs. There is a strong financial incentive therefore to use telematics to spot drivers who regularly brake, corner or accelerate harshly, and intervene before this behaviour results in costly collisions or bigger bills.
Don’t rely on self-reporting alone
Researchers at Loughborough University in the UK compared what drivers said about their own driving against a hard dataset. Data from 96 drivers, 131,462 trips, 1.46 million kilometres, and over 32,000 hours on the road was used to assess how robust self-reporting was. Questionnaire answers tracked actual speeding only loosely; road type, weather, daylight, posted limits and surrounding traffic explained far more of what happened. However, in many cases the drivers’ own accounts of their speed generally didn’t line up with what was recorded. In fact, drivers who said they never exceeded the speed limit actually spent 42% of their driving time above it. That gap isn’t dishonesty; it likely reflects that people are simply poor at recalling habitual behaviour. Driver questionnaires should be treated as a starting point, but they shouldn’t be the primary way a fleet identifies risk.
Use scores as a starting point, not a verdict
A 2025 academic review of driver-behaviour scoring systems assessed both the potential and the limitations of AI-driven driver assessment. The research found that while AI can help to recognise patterns such as speeding or harsh braking, and predict potential maintenance issues by highlighting them early, the authors also flagged that generalising a model trained on one vehicle type, route or driving condition to another remains one of the field’s biggest challenges. Furthermore, the authors also noted that data privacy is an ongoing concern as these systems scale.
Putting the research into practice.
A 2025 conference paper by Suryakant Kaushik proposes a fleet-management framework that brings routing, maintenance, driver analytics and sustainability data together into one system.
A Fleet Manager’s checklist
- Don’t attempt to tackle all issues at once, pick one cost problem to focus on.
- Ensure the peer groups are actually comparable so the comparisons will actually mean something.
- Map incidents by time and location to identify whether particular routes, delivery windows, customer sites or vehicle-and-job combinations are increasing costs that look at first glance, like a driver problem.
- Cross-reference driver-behaviour data with maintenance records to determine whether harsh braking, rapid acceleration or speeding are actually increasing wear, repairs or downtime.
- Finally, track performance over time to see whether driver coaching, better route planning or different vehicle allocation produce measurable gains in fuel efficiency, component longevity, vehicle uptime and overall availability. If they’re not, some other area needs to be assessed.
Conclusion
Research consistently highlights that driver behaviour is a genuine fleet cost variable. But driver performance cannot be meaningfully assessed without considering the conditions in which vehicles operate. Route design, delivery pressures, vehicle allocation, payload and traffic conditions all matter.
Used properly, telematics does more than rank people. It gives fleets an earlier, sharper read on where safety and cost problems are actually originating from allowing for earlier, more targeted safety and cost interventions. However, the important caveat is that results must be interpreted fairly and in context.
Sources:
1 – Zhang, Z., Demir, E., Mason, R. and Cairano-Gilfedder, C. (2023), “Understanding freight drivers’ behavior and the impact on vehicles’ fuel consumption and CO2e emissions in road freight transport”, Operational Research, 23(4), doi: 10.1007/s12351-023-00798-2.
2 – Xiang, J., Ghaffarpasand, O. and Pope, F. D. (2024), “Mapping urban mobility using vehicle telematics to understand driving behaviour”, Scientific Reports, 14, article 3271, doi: 10.1038/s41598-024-53717-6.
3 – DriveTech (2020), The Impact of Driver Behaviour on Vehicle Running Costs.
4 – Thomas, P., Welsh, R., Morris, A. and Reed, S. (2025), “Validating self-reported driving behaviours as determinants of real-world driving speeds”, Ergonomics, 68(8), pp. 1192–1206, doi: 10.1080/00140139.2024.2395419.
5 – Shirole, V., Shahade, A. K. and Deshmukh, P. V. (2025), “A comprehensive review on data-driven driver behaviour scoring in vehicles: technologies, challenges and future directions”, Discover Artificial Intelligence, 5, article 26, doi: 10.1007/s44163-025-00244-6.
6 – Kaushik, S. (2025), “Driving Innovation in Fleet Management: An Integrated Data-Driven Framework for Operational Excellence and Sustainability”, Proceedings of the 14th International Conference on Data Science, Technology and Applications (DATA 2025), pp. 500–507.