Fuel Theft Savings Calculator
Before you buy tank sensors for a fleet, somebody will ask what they are worth. This calculator sets the value of the fuel you cannot currently account for against what a detection programme costs to run, and returns a net figure and a payback period.
Your numbers
The vehicles that would actually get a probe, not the whole fleet.
Total litres per vehicle per year from your card and bowser records.
Probe, installation labour and calibration. Use a real quote from your installer rather than the figure sitting in this field.
Platform subscription and data. Use your own quoted price.
Result
Net saving in year one
14,456
Recovered fuel value minus year-one programme cost across 40 vehicles.
- Annual fleet fuel spend
- 1,612,000
- Value of unaccounted fuel
- 48,360
- Value you expect to recover
- 29,016
- Hardware cost, one-off
- 8,800
- Subscription cost per year
- 5,760
- Year-one programme cost
- 14,560
- Payback
- 6.0 months
The formula
- Fleet litres = Fuel bought per vehicle × Vehicles to be fitted
- Annual fleet fuel spend = Fleet litres × Fuel price
- Value of unaccounted fuel = Annual fleet fuel spend × (Share of fuel you cannot account for ÷ 100)
- Value you expect to recover = Value of unaccounted fuel × (Share of that gap you expect to close ÷ 100)
- Hardware cost, one-off = Hardware and fitting per vehicle × Vehicles to be fitted
- Subscription cost per year = Software per vehicle × 12 × Vehicles to be fitted
- Year-one programme cost = Hardware cost, one-off + Subscription cost per year
- Net saving in year one = Value you expect to recover − Year-one programme cost
- Payback = Year-one programme cost ÷ (Value you expect to recover ÷ 12)
Assumptions and limits
- Every default in this tool is a stand-in, including the unaccounted share and the recoverable share. Neither is a benchmark and neither came from research. Replace both with your own figures or the output is fiction with a currency symbol.
- Recovery is assumed to happen evenly from day one. It will not. Sensors have to be fitted, calibrated and validated before a single alert means anything, so the real payback is later than the arithmetic says.
- The model treats the whole gap as recoverable in principle. Some of it is not theft at all: it is calibration drift, an over-reading bowser, a consumption model that runs optimistic, or a driver filling a jerry can with permission nobody wrote down.
- Deterrence is not modelled, and it is often the larger effect. Losses tend to fall once people know the tank is watched, which shows up as a drop in the gap rather than as a stack of caught events. This calculator has no way to represent that.
- Your internal time is excluded. Somebody has to work the alerts, resolve the events and have the conversations, and that is a real cost this model does not carry.
- Year one only. Hardware is one-off and subscription recurs, so year two economics look considerably better than the headline figure here.
Start with the gap, not with the theft
The number you can defend is not how much fuel is being stolen. Nobody knows that. The number you can defend is the difference between litres purchased and litres your consumption model says should have been burned, over a long enough window that noise cancels out. That residual is the gap, and it is the only honest starting point for this calculation.
It is also uncomfortable, because a large gap does not prove wrongdoing and a small gap does not prove innocence. What it does is size the uncertainty. If your books balance to within a rounding error, tank sensors are solving a problem you do not have. If they do not balance, you at least know what the unknown is worth per year.
Why the recoverable share is where the model lives or dies
Multiply the gap by 100 percent and the business case looks wonderful. It is also wrong. A meaningful share of any unaccounted fuel turns out to be measurement error once you start looking: a bowser that dispenses 205 litres and records 200, a probe calibrated on the wrong tank geometry, a consumption baseline built in summer and applied in January.
Those discoveries are worth something too, but they are worth it once, not annually. So be pessimistic in that field. A business case that survives a conservative recoverable share is one you can take to a finance director. A business case that only works at ninety percent recovery is a business case that will be remembered for the wrong reasons.
What sensors change, and what they do not
A calibrated tank probe converts a monthly argument into a timestamped event. It shows a forty-litre drop at 02:40 at a location that is not a depot with the ignition off. That is not a suspicion; it is a record with a coordinate. The value is not really the litres recovered from that one event, it is that the argument about averages ends.
What sensors do not do is manage the consequence. Somebody still has to review the alert, decide whether it was theft or a fault, and act. Fleets that fit hardware and never staff the follow-up get the deterrence effect for a while and then get nothing, because the alerts pile up unread and the litres come back. The programme cost in this model is hardware and software. The part that actually determines whether it works is not in any field on this page.
Frequently asked questions
Take a long window, ideally a full quarter. Add every litre bought against the vehicle from cards, depot bowser and manual slips. Take the distance from odometer readings at both ends, and multiply by your best real-world consumption rate to get expected litres. The difference divided by litres bought is your gap. Do it per vehicle: a fleet-wide figure averages away the one truck that is the whole problem.
Probably not all of it, and assuming so is how fleets damage relationships with good drivers. The list of innocent explanations is long: an over-dispensing pump, a probe calibrated against the wrong tank profile, a consumption baseline that never accounted for the winter, a route that got hillier when a bridge closed. Sensors are useful precisely because they let you eliminate those before anyone is accused.
Because we would have to invent it. There is no defensible average across regions, fuel prices, security arrangements, route types and depot practices, and a number presented as typical would be doing persuasion rather than arithmetic. You have the data to measure your own gap. That number is worth more than any benchmark, because it is true.
Yes, and deliberately so, so that you can see the assumption. It divides year-one cost by an even monthly recovery, which assumes recovery starts immediately and runs flat. In practice the first weeks go on fitting, calibration and baselining, and early alerts get resolved as sensor faults rather than theft. Add two to three months to whatever this returns before you commit to a date in front of anyone.
No, and starting with the whole fleet is usually the expensive way to find out. Fit the vehicles where the gap is largest and the exposure is highest: long routes, overnight parking away from a depot, high-value fuel loads. If the pilot closes a measurable gap on those, the case for the rest writes itself with your own evidence. If it does not, you learned that cheaply.
Stop estimating. Measure it.
Fuel Theft Savings Calculator gives you the arithmetic. Fleet Fuel Management Software gives you the live numbers from your own fleet.