How Custom Fuel Management Software Cut a Fleet's Costs by 35%

The frustrating thing about fleet fuel costs is that everyone knows they're too high. The data to prove it — and to identify exactly where the waste is occurring — almost never exists.

That was the situation at EMAYYAM INFOTECH, an IT services company in Coimbatore managing a fleet of vehicles across their field operations. Their Head of Business Development, Jey C., approached us with a specific problem: fuel costs were running significantly above what the fleet size and operational requirements should justify, but without the data infrastructure to understand the root cause, there was nothing to act on.

The Problem Was Visibility, Not Discipline

Fleet fuel management problems almost always start the same way: fuel is being purchased using company fuel cards, paper logs are being kept inconsistently, and someone in finance is looking at a line item on the P&L that's too large but has no way to drill into it.

At EMAYYAM, three specific problems were driving the cost overrun — but they only became visible after the tracking system was in place:

Vehicle-level inefficiency. Several vehicles were consistently consuming more fuel than their benchmarks. This was a maintenance signal — poor tyre pressure, engine issues, sub-optimal running conditions — but without per-vehicle consumption data, it was invisible.

Sub-optimal routing. Drivers were taking familiar routes rather than efficient ones. Without route-level consumption data, there was no way to identify which routes had the highest variance or which drivers were consistently running longer distances for similar jobs.

Procurement leakage. Without automated reconciliation between fuel card statements and vehicle usage records, small discrepancies — potential indicators of misuse — were extremely difficult to detect in a manual system.

What We Built and Why

The scope was deliberate: a custom fuel management system, not an off-the-shelf fleet management platform. We've seen clients buy expensive fleet management SaaS only to find that 80% of the features don't apply to their operation, the ERP integration doesn't match their data model, and the reporting doesn't answer the specific questions their operations team needs to ask.

A custom build for a focused problem is often faster and cheaper than configuring a platform, and it integrates cleanly with the systems the client already uses rather than creating a parallel data silo.

The system we built in six weeks had four modules:

Vehicle and driver registry. Each vehicle registered with its specification, fuel type, and expected consumption benchmarks. Each driver associated with their typically assigned vehicles. This is the foundation — without clean reference data, anomaly detection produces false positives that erode trust in the system.

Fuel transaction log. Replacing the paper-based system with structured digital records. For supported fuel card providers, transactions were imported automatically via API. For others, they were entered via the web app. Each transaction was validated against the vehicle's expected consumption rate — transactions outside expected parameters were flagged automatically for review.

Analytics dashboard. Fuel cost per vehicle, per route category, and per driver in real time. Month-on-month trend analysis. And the most operationally useful feature: outlier detection — a visual display of vehicles whose consumption was consistently above fleet average, with drill-down capability to see the specific transactions driving the deviation.

ERP integration. A scheduled batch export at the end of each day generated a structured transaction file in the format required by EMAYYAM's existing ERP, imported automatically into the cost accounting module. No manual data re-entry. Fuel cost data available in the ERP with a maximum one-day lag.

The Anomaly Detection Logic

The consumption anomaly detection algorithm compared each transaction against a rolling baseline for that vehicle: the median consumption per kilometre over the preceding 30 days. Transactions more than 1.5 standard deviations above or below the baseline were flagged.

We also implemented a driver attribution model: because vehicles were often driven by multiple drivers, the anomaly detection cross-referenced driver assignment data to distinguish vehicle-level issues (consistent deviation regardless of driver) from driver behaviour issues (deviation concentrated in specific driver-vehicle combinations). This distinction matters for the intervention — a vehicle-level issue requires maintenance; a driver-level issue requires a different kind of conversation.

The statistical threshold was deliberately conservative. A 1.5 standard deviation cutoff means fewer flags but higher signal quality. A more aggressive threshold would have generated more flags but eroded trust in the system when operations staff investigated and found legitimate explanations.

The Results

Within the first months of deployment, the system surfaced findings that were immediately actionable:

• Several vehicles consistently consuming above benchmark — maintenance issues confirmed and addressed
• Route patterns showing higher-than-expected variance — dispatch routing adjusted
• A small number of fuel card transactions flagged as inconsistent with vehicle usage records — investigated and resolved

The aggregate result: fuel costs fell 35% compared to the pre-deployment baseline. This substantially exceeded EMAYYAM's own estimate of what was achievable. The improvement came from both the direct interventions enabled by the new visibility and the secondary effect of simply having a tracking system in place — awareness of monitoring changes behaviour.

Vehicle maintenance costs fell 15% as condition-based scheduling replaced calendar-based scheduling. Vehicles showing consumption patterns associated with developing mechanical issues were serviced proactively rather than waiting for a breakdown in the field.

"Srijith and team were able to grasp the requirements and come up with a solution immediately. We were able to cut down on our fuel expenses up to 35%. Thanks to Srijith and team for making our task easier at an affordable cost." — Jey C., Head of Business Development, EMAYYAM INFOTECH

The Broader Lesson for Fleet-Intensive Businesses

The return on investment from fuel and maintenance savings alone significantly exceeded the development cost within the first year of operation. This is not unusual for operational data infrastructure: the value of visibility is high in any asset-intensive business, because the waste is already happening — you're just paying for it without knowing exactly where it's going.

If your fleet, manufacturing, or field operations have a cost line that's too high but you don't have the data infrastructure to understand why, that's the problem worth solving first. The interventions almost always follow from the visibility — you don't need to know what to fix before you build the system; the system tells you what to fix.

We build this kind of operational software — fuel management, fleet ERP, field service management, manufacturing analytics — for companies across India, the Middle East, and Southeast Asia. Book a 15-minute call if you want to talk through what a custom system would look like for your operation.