Spotting fuel loss: why a vehicle is compared to itself, not to a norm

Every fleet manager who has ever suspected fuel loss has tried the same two things: a consumption norm from a manual, and a check after every refuelling. Both fail, and they fail for reasons worth understanding before you build a control process on top of them.

Why a per-trip check is noise

Drivers do not fill the tank in proportion to the trip they are on. A full tank logged on a five-kilometre hop looks like catastrophic consumption; three short trips after one fill look like a vehicle running on air. A per-trip fuel-efficiency alert therefore fires constantly and is almost always wrong.

An earlier version of Depot 17 had exactly that check. It produced so many false alarms that people stopped reading the alerts entirely — which is worse than having no check at all, because it also buries the real one. It was removed.

Why a manual norm is the wrong yardstick

A manufacturer figure describes a new vehicle, on a test cycle, at a steady speed, unladen. It says nothing about a nine-year-old van, loaded, on broken roads, in winter, doing short urban runs. Judged against the book, half a real fleet is permanently "over norm" — and once everything is flagged, nothing is.

The useful comparison is not the vehicle against a book. It is the vehicle against itself, last quarter.

The check that actually works

Consumption is only meaningful over a period and a whole vehicle, so that is the unit. Each vehicle is measured over a rolling 30-day window and compared with its own baseline: the 90 days before that window, with no overlap between the two.

The rolling window matters more than it sounds. A check that measures "since the start of the month" is blind on the first of every month and, on the second, judges a fleet on one day of driving. A rolling window always has a full period behind it.

Three independent judgements come out of it, all in litres (or kWh) per 100 km, the unit of the waybill register:

• Implausible — the ratio is outside what the vehicle could physically achieve: roughly 3.3 to 50 litres per 100 km for a combustion engine, 6.7 to 67 kWh per 100 km for an electric one. This is almost never fuel loss. It is a data problem: a missing odometer reading, litres typed into the cost field, a refuelling logged twice. Fix the entry.
• Deviation — the ratio is entirely plausible but sits 25 per cent or more away from that vehicle own baseline. This is the one that matters. A fuel leak, a dragging brake, a failing injector, a change in how the vehicle is being driven — or fuel going somewhere it should not.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.
• Over the norm: when the vehicle card holds a consumption norm (the one the waybill register writes fuel off by), a window 50 per cent or more above it is flagged. The margin is wide on purpose: a norm is a round figure, and a vehicle a few per cent over it is not news.

The guards that keep it quiet

A check that cries wolf is worthless, so three rules hold it back. A window with less than about 150 km or 15 litres behind it is too thin to judge and is skipped. The baseline has to clear the same floors, so a vehicle that has only just entered service is never accused of anything. And hybrids are excluded outright — two energy sources and one ratio produce a number with no meaning.

The result is a monthly digest that is usually short and occasionally worth acting on, which is the only kind anybody reads.

What to do with a deviation

Take it as a question, not a verdict. In order of likelihood: check the odometer readings around the window for a missed trip; check whether the vehicle changed duty — city runs consume more than highway ones and a rota change alone can move the figure; check the service history for something mechanical. If none of those explain it, you have a fuel question, and now you have it with dates, quantities and odometer readings attached rather than as a suspicion.

That last part is the real value. Discussing a possible fuel loss with a driver goes very differently when you can point at a specific 30-day window, the vehicle own previous quarter, and the refuelling rows behind both.

What it is not

This is arithmetic over what your team recorded, not telemetry. There is no fuel-level sensor and no live vehicle tracking behind it. If a driver never logs a refuelling stop, no calculation can see it — which is why the odometer at the pump is worth insisting on, and why the check gets sharper the more disciplined the entries are.

Depot 17 records trips and refuelling from the driver phone, offline when needed, and turns those records into monthly reports and a consumption digest. Register your organisation and give it a quarter of data to work with.