Fuel is the largest controllable cost in most Kenyan transport businesses, and the one operators have the least honest information about. The reason is structural: EPRA publishes maximum retail pump prices monthly, so the price per litre is broadly the same wherever your vehicles fill. You cannot negotiate your way out of the problem. The only variable you genuinely control is litres — and litres are exactly what most fleets cannot account for.
This guide is vendor-neutral on hardware. It describes how the measurement actually works so you can judge any supplier's claims, including ours.
Every fuel monitoring setup is built from some combination of three inputs. Most disappointing deployments come from an operator buying one of them and expecting it to do the job of all three.
Receipts are necessary and insufficient. They establish what you paid for. They establish nothing about where it went. A fleet running on receipts alone is measuring its own procurement, not its fuel.
This is where a lot of fuel projects quietly fail. CAN telematics is genuinely valuable — an odometer reading nobody can adjust is worth a great deal on its own — but the assumption that any modern truck will hand over a clean tank level does not survive contact with a mixed Kenyan fleet. Ask a supplier to demonstrate fuel level on your vehicle makes before you build a programme on it.
A probe fitted into the tank sidesteps the whole question of what the vehicle is willing to tell you. It costs more to install, it has to be calibrated properly against the tank it sits in, and it is the option most likely to still be working across a fleet of different makes and ages.
Fuel accountability without a fuel screen in the cab.
Kora records every fill against the vehicle with odometer, cost per litre, station, and receipt, then compares litres to the distance the vehicle actually covered. Anomalies go to a review queue in the workspace — never to the driver's phone.
Underneath every fuel monitoring product is one piece of arithmetic. It is not complicated, which is why it is worth understanding before you pay anyone for a dashboard that performs it:
The critical word in that list is baseline. A vehicle's own historical consumption on its own routes with its own loads is the only honest comparison. A brochure figure will flag half your fleet as anomalous and train everyone to ignore the alerts within a fortnight. Establish the baseline from real running, and expect it to differ between a loaded Mombasa run and an empty return.
Here is the thing most demos will not raise: a fuel tank on a Kenyan road is not a still container. Fuel moves. Levels swing on cambers, potholes, roundabouts, and gradients. Tank senders themselves differ — some smooth their output heavily before reporting, others report the raw swing — so two trucks in the same yard can produce very differently behaved data from the same event.
This is why a system that fires an alert on a single reading is worse than useless: it will report a theft every time a truck takes a roundabout, and within a month nobody will open the notifications. A system worth paying for judges a window of readings — sustained change over time, sanity-checked against whether the vehicle was moving, running, or stationary — and only then raises something for a human.
Ask any vendor directly how they handle sender noise and stationary-versus-moving context. The answer tells you whether you are buying a monitoring system or a chart.
A useful fuel programme produces a short, specific queue of things a human should look at — not a wall of readings. In practice that means a small set of named conditions:
Every one of those is a question, not an accusation. Vehicles get topped up before a long run, senders fail, receipts get lost, and a genuinely heavy load burns more. The value is that the question gets asked within days, against a specific vehicle and a specific fill, instead of surfacing as an unexplained monthly total nobody can reconstruct. For the human side of that conversation, see the practical guide to stopping diesel theft in a Kenyan fleet.
Most fleet products hand fuel capture to the driver: log the fill, photograph the receipt, done. It is convenient, and it quietly destroys the control. You have asked the person whose consumption is being measured to own the measurement, and you have told them precisely what is being watched.
The stronger arrangement is straightforward: fills and receipts are captured by the office against the vehicle, and the vehicle's own telemetry provides an independent reference that nobody in the cab has to be aware of. The driver's job is to drive. The record belongs to the operation. That separation is worth more than any single feature in a fuel product.
It also changes what you learn. When the measurement is visible to the person being measured, you find out what they are willing to report. When it is not, you find out what actually happened — and the gap between those two is the entire subject of the hidden fuel leaks in a transport business.
Four questions, in order. What does this read on my vehicle makes — verified on one of my trucks, not a demo unit? Where does distance come from, and can a driver influence it? How does the system separate real level changes from road noise? And when it flags something, who reviews it, and what happens next?
A supplier who answers all four clearly is selling you a fuel programme. A supplier who leads with a live tank-level graph is selling you a screen. If you are still choosing hardware, the guide to vehicle trackers in Kenya covers what each class of device does before fuel enters the picture.
A fuel monitoring system tracks how much fuel a vehicle receives and how much it consumes, then compares the two against the distance covered. In practice it combines three inputs: fill records showing litres purchased, distance and engine hours from the vehicle, and a tank level reading from either the vehicle's own data or a sensor fitted in the tank. Discrepancies between what was bought and what the work required are flagged for a person to review.
It depends on your vehicles, and it must be verified rather than assumed. CAN data reliably supplies odometer and engine hours on most modern trucks, which is the harder half of the consumption calculation. Usable tank level over CAN is much less consistent, and several truck families common in Kenyan fleets publish little or none. For mixed fleets, an in-tank fuel level sensor is usually the dependable path for level, with CAN supplying distance alongside it.
It does not detect theft directly — it detects discrepancies that theft produces. A level drop while the vehicle is stationary and not running, litres claimed that the distance covered cannot justify, a fill larger than the tank can hold, or repeated fills with no mileage between them all become visible. Each one is a flag for a human to investigate, not a verdict. No system makes a fleet theft-proof; the point is to make loss visible quickly and expensive to attempt.
Because the price is largely set for you. EPRA publishes maximum retail pump prices for diesel, petrol, and kerosene monthly, so the per-litre cost is broadly similar wherever your vehicles fill. Negotiating a few shillings is a small lever compared with knowing whether every litre you paid for actually did work. Litres consumed is the variable an operator genuinely controls.
It is a weak control. Asking the person whose consumption is being measured to also own the measurement removes most of the value, and it tells the driver exactly what is being watched. A stronger arrangement is for fills and receipts to be captured by the office against the vehicle, with the vehicle's own telemetry as an independent reference. The driver's job is to drive; the record belongs to the operation.
Accurate enough to separate a real loss from normal variation, which is a lower bar than perfect litre-level precision. Fuel moves in a tank on rough roads, and a single reading can be misleading in either direction. What matters is whether the system judges trends over a window rather than reacting to individual readings, and whether the flags it raises turn out to be worth investigating. A system that cries wolf gets ignored, and an ignored system controls nothing.
Every litre, accounted for
Kora records every fill against the vehicle, compares litres to the distance the vehicle really covered, and routes excess fills, mileage mismatches, and suspected siphoning to a review queue with a named decision — not a chart nobody opens.
