Volodymyr Zavadko, delivery director, transportation at Intellias, explores why the biggest challenge facing fleets today is not collecting more data, but turning it into better operational decisions.

Ask any fleet manager whether they have enough data, and the answer is almost always yes. Connected vehicles constantly generate an abundance of telemetry, yet many organisations still struggle to translate that information into measurable operational or financial benefits.

Greater visibility has not automatically led to better outcomes. While fleet managers have access to more information than ever before, much of it remains disconnected from the decisions that influence cost, efficiency and vehicle uptime.

In many cases, the problem is not a lack of data but a lack of integration. Valuable information is often spread across telematics platforms, maintenance systems, fuel cards, charging networks and operational software, making it difficult to turn insight into action.

As a result, fleets can find themselves collecting vast amounts of information without fully benefiting from it. There’s a very real opportunity here for fleet managers who seek to close the gap between what vehicles are reporting and the decisions being made about how those vehicles are maintained, deployed and managed throughout their lifecycle.

The problem isn’t data, it’s disconnected systems

Most fleets today manage increasingly complex and diverse vehicle parcs, with legacy internal combustion vehicles operating alongside light commercial vehicles and a growing share of electric vehicles. Each brings its own telemetry hardware – aftermarket OBD-II dongles, video telematics units, factory-installed OEM systems – none of which is initially designed to speak the same language as the others.

The result is that a fleet’s data lives in four or five different silos that were never built to communicate: telematics in one portal, fuel or energy transactions in another, maintenance and leasing records somewhere else again. Even where the data is technically available, it often surfaces in the wrong place. For example an efficiency report that lands as a monthly PDF will likely change nothing, because by the time anyone reads it, the decision it could have informed has already been made.

The fix is putting the right insight in front of the right person at the point they’re making a call like a route recommendation on a dispatcher’s screen the moment they assign a job, a maintenance alert in a workshop planner’s queue when a fault code appears, or a charging recommendation before an EV is plugged in.

The greatest value here comes from embedding intelligence into operational workflows, rather than creating another reporting layer beside them.

The biggest gains come when data shapes day-to-day decisions

Once data is structured and connected to a workflow, three areas consistently deliver returns.

1. Predictive maintenance: Fixed service intervals lead to two failure modes which are over-servicing healthy components and missing the ones about to fail. It’s the difference between an annual check-up and a doctor watching your vitals in real time.

Capturing real-time diagnostic trouble codes and sensor data allows fleets to build a live model of each vehicle’s condition, increasingly through digital twin technologies that create virtual representations of critical vehicle systems. These models can identify patterns of wear before they result in breakdowns, helping operators reduce unplanned downtime, improve vehicle availability and protect residual asset values.

2. Smarter utilisation and routing: Combining location intelligence with live traffic data allows routes to adjust as conditions change rather than following a static plan. Fleets that get this right can improve route efficiency by up to 20%, which in turn allows them to operate fewer vehicles – often 15-30% fewer – while maintaining service levels.

The biggest savings for fleets often don’t come from buying new assets but from making better use of the ones they already have. Better visibility into utilisation can reveal underused vehicles, reduce unnecessary fleet growth and improve overall transport capacity planning.

3. EV charging and grid participation: EVs add variables that ICE fleets never had to manage or account for such as state of charge, battery degradation, charging duration and tariff timing. Aligning dispatch schedules with off-peak charging windows can reduce energy costs and help avoid peak-demand penalties.

Go a step further and idle EVs can be pooled into virtual power plant (VPP) schemes, allowing battery capacity to support the wider electricity grid during periods of high demand. In some markets, this is already creating new recurring revenue opportunities for commercial fleet operators, turning vehicles from purely operational assets into potential energy assets too.

AI is only ever as good as the data underneath it

There’s real momentum behind AI for range prediction, demand forecasting and anomaly detection, and it’s very much justified. But a model is only as reliable as the data layer it sits on. Feed noisy, unparsed logs into a machine learning pipeline and the output is simply more false alerts.

Getting the foundation right is critical, and it largely hangs on treating fleet data as infrastructure. That means implementing automated pipelines to clean and normalise raw signals before they reach an algorithm, building cloud architectures capable of handling years of historical telemetry alongside real-time data, and establishing governance frameworks that meet automotive-grade standards such as ASPICE and TISAX.

It also requires the use of common data models, such as COVESA VSS, so information from different OEMs and hardware vendors can be compared and analysed consistently.

Just as importantly, the most successful deployments keep experienced fleet professionals in the loop. AI works best when it supports operational decision-making rather than attempting to replace it. Recommendations need to land inside the ERP or transport management system a planner already uses, not a separate dashboard they have to remember to check.

Skip this foundational step and the technology quickly loses credibility. The model may continue to generate recommendations, but if fleet operators do not trust the outputs, they are unlikely to act on them.

Those seeing results are already proving the model

The benefits are already highly visible in organisations that have shifted their focus beyond data collection and more into operational decision-making.

DHL, for example, combined video telematics with driver coaching to reduce accidents by 26% and cut accident-related costs by almost half. At Denver International Airport, bringing together asset utilisation data from multiple departments helped avoid $400,000 in vehicle procurement costs by identifying opportunities to make better use of existing resources.

Meanwhile, DKV Mobility illustrates how connecting mobility data can improve cost visibility at scale. Its platform brings together transaction data from refuelling, EV charging, tolls and vehicle services, giving fleet operators a more consolidated view of costs and consumption.

In each case, the gains came not from collecting more information, but from using existing data to support better operational decisions.

Better data decisions will define the winners of fleet performance

The technology to capture fleet telemetry is mature and widely available – that piece of the jigsaw is largely solved. What separates fleets that turn data into measurable value from those still generating reports nobody reads is whether the insight reaches the person who needs it, in time to act on it, at every stage of the fleet lifecycle.

For many operators, the biggest gains are to be found in making better use of the data they already have, and not from collecting more of. Certainly, the fleets seeing the most progressive results are those using information to shape everyday decisions around maintenance, utilisation, routing and charging.

At the end of the day, data only creates value when it changes what happens next.