Every fuel station loses product on paper. That is normal, and the amount is usually small. The problem starts when the number stops being small and nobody can explain where the fuel went. Fuel variance is the gap between what your books say you should have and what your tanks actually hold. On a station moving 200,000 litres a month, a variance of 0.25 percent is 500 litres nobody paid for. Over a year that is 6,000 litres, or about $6,000 at a pump price of a dollar a litre. That is not a rounding error. It is a salary, a new dispenser, or the maintenance budget you keep postponing. Most operators treat variance as a theft problem. In practice theft is one of the less common causes, and the easiest to rule out once you have daily data. The bigger contributors are worn meters, wrong tank charts, temperature, and plain paperwork. None of them make headlines. All of them cost money every month. This guide is for station owners and operations managers who already have numbers in front of them and want to know which ones matter.
What fuel variance actually is
Variance is a book-to-tank comparison, nothing more. You take the opening stock, add every delivery, subtract every litre the dispensers recorded, and compare the result with the physical stock your tank gauge reports. The difference is your variance for that period, usually expressed in litres and as a percentage of throughput.
- Book stock = opening stock + deliveries − recorded sales
- Variance = book stock − physical stock
- Variance percentage = variance ÷ throughput × 100
The sign matters. A positive variance means the tank holds more than the books expect. A negative variance means product has disappeared. Both are worth watching, and a station that swings between the two is telling you something different from a station that loses steadily in one direction.

Why a fraction of a percent is real money
Percentages hide the size of the problem. Here is the same station at 200,000 litres a month, at a pump price of one dollar per litre.

| Variance | Per month | Per year | Annual value |
|---|---|---|---|
| 0.10% | 200 L | 2,400 L | $2,400 |
| 0.25% | 500 L | 6,000 L | $6,000 |
| 0.50% | 1,000 L | 12,000 L | $12,000 |
| 1.00% | 2,000 L | 24,000 L | $24,000 |
Many operators set their investigation threshold at 0.5 percent of monthly throughput. That is a reasonable place to put the alarm, but it is not a target. If your cumulative variance drifts past 0.25 percent for three months running, you already have a problem worth a morning of work.
The four places fuel goes

Measurement error
This is the category operators underrate and the one that usually explains a variance that will not move no matter how many people you interview.
- Meter slippage. Positive displacement meters wear. As internal clearances open up, product slips past the measuring chamber uncounted, and the dispenser under-registers. The station gives away fuel it never bills. A worn meter can drift 0.2 to 0.5 percent before anyone notices, because every individual fill still looks perfectly normal.
- Tank charts and gauge drift. The chart converts a liquid level into a volume. Real tanks are not the perfect rectangles the chart assumes, and tilt, baffles and sediment all distort the curve. A chart built for a different tank shape can be wrong by tens of litres at some levels and almost right at others. This is one of the most common single causes of a stubborn variance.
- Temperature. Fuel expands as it warms. Gasoline changes volume by roughly 0.1 percent per degree Celsius. A tank sitting at 22°C in the afternoon and 12°C at dawn swings about one percent across the day. If you dip at different times on different days, part of what you are measuring is the weather.
- Legal metering tolerance. Dispensers are allowed a small error under the standards they are approved to. OIML R117 permits ±0.5 percent on the volume delivered. Two dispensers sitting at opposite ends of that tolerance will generate variance even when both are working as designed.
Physical loss
A slow leak is both a product loss and an environmental liability, which is why regulators such as the US Environmental Protection Agency require leak detection on underground storage tanks. Water arrives the same way and does the same kind of damage: it takes up volume, some float gauges count it as product, and a water level that creeps upward points at a bad fill cover, a damaged gasket, or groundwater finding its way in. Evaporation is the mildest of the three. It is modest on a busy tank, noticeable on a low-turnover grade, and worse in hot climates where vapour recovery is missing or broken.
Paperwork and process
This is the unglamorous category, and on many stations it is the largest. Fuel leaves the site without ever being billed, and the tank is the only witness.
- Deliveries that were not fully discharged. The compartment was not emptied, the driver left product in the truck, and the tank holds less than the delivery note claims.
- Drive-offs and credit sales that never made it into the system.
- Litres used for testing, filter changes, generator runs, and the workshop vehicle.
- Product moved between tanks without a record.
Theft
It deserves a clear-eyed look rather than a shrug. What matters is that theft has a signature. It concentrates: one shift, one attendant, one dispenser, one nozzle. It often comes with repeated small overrides, manual pump starts, or dispensing outside business hours. It frequently stops the week a particular person leaves. If your variance is spread evenly across every shift, every day, always in the same direction, it is almost never theft. That pattern points at the equipment.
Reading the pattern before you act
The fastest way to cut the list of suspects is to look at how the fuel variance behaves over time, not how large it is.

| What you see | Most likely cause | First check |
|---|---|---|
| Small, steady loss every day, all shifts | Meter slippage or tank chart error | Meter calibration, tank chart against a fresh calibration |
| Variance swings between gain and loss | Temperature, or inconsistent dip timing | Dip at the same hour, correct to a reference temperature |
| Loss appears only after deliveries | Short delivery or transfer not recorded | Delivery notes, truck meter, discharge records |
| Loss concentrated on one shift or one person | Theft or unrecorded sales | Transaction log for that shift, per-nozzle totals |
| One tank drifts, others are stable | Tank-specific leak, water, or gauge fault | Water level, overnight stock test on that tank alone |
| Large one-off loss | Spill, overfill, or a big unrecorded withdrawal | Delivery day records and tank gauge alarms |
A daily routine that takes fifteen minutes
- Read the tanks at the same time every day. Early morning, before the temperature climbs, gives you the most comparable numbers.
- Reconcile deliveries separately. Compare the delivery note against what the tank actually gained, and do it before the fuel gets sold.
- Work at a reference temperature. Raw gauge litres at 30°C are not the same litres your invoices are based on. Correct them.
- Track cumulative fuel variance, not just daily. Daily figures are noisy. The thirty-day trend is the signal, and the trend is what tells you whether a fix worked.
- Suspect the equipment before the people. Calibration is cheaper than an investigation, and it is the answer more often than most owners expect.
What changes when the station is instrumented
Automatic tank gauging and forecourt control do not make fuel variance disappear. They make it attributable, which is the part that actually saves money. The difference between a monthly dip and a continuous reading is the difference between finding a leak in November and finding it on Tuesday.

- i-Level ATG keeps continuous watch on product level, water and temperature in every tank, maintains stock automatically, and raises an alarm when a tank drops outside its normal pattern.
- i-LS magnetostrictive level probe is the sensor underneath that measurement. Probe accuracy is the ceiling on everything above it, and a float gauge on a tilted tank will never give you a clean reconciliation.
- i-Terminal links dispensers, tanks and back office, so per-nozzle totals arrive in the system without anyone reading a mechanical counter and typing it in.
- i-Console consolidates stock, sales and variance across stations in one place, which turns a monthly argument about one site into a portfolio view where the outliers stand out on their own.
- i-Reader ties every litre to a card, vehicle or driver at the dispenser. It removes most of the unrecorded-sale category at the source instead of explaining it afterwards.
- i-Pay outdoor payment terminal captures payment at the pump together with the transaction, so the sale and the litre stay attached to each other.
- i-Truck covers the delivery side: what left the depot, what arrived, and what was actually discharged into which tank. Short deliveries stop being a monthly surprise.
The point of the list is not to buy all of it. Each item closes a specific hole. A station losing fuel to short deliveries needs delivery tracking, not a new tank gauge. A station losing it to card fraud needs attribution at the dispenser, not a better chart. Start from the pattern in the table above and fix the cause you can actually see.
What the numbers look like afterwards
The honest answer is that fuel variance does not go to zero. Meters wear, tanks are imperfect, and fuel is a liquid that moves. What changes is the size of the surprises and how long they last. When reconciliation runs daily instead of monthly, the feedback loop shrinks from thirty days to one. A drifting meter or a bad chart that would previously have run for a year and been written off at inventory time now shows up in the first week. On the numbers above, catching a 0.25 percent problem three months earlier is worth about 1,500 litres on a single station. Across a network, that is the difference between an operating cost and a line item you can actually control.
Frequently asked questions
What is a normal fuel variance for a petrol station?
Is fuel variance always caused by theft?
Can temperature really cause fuel variance?
What is the difference between fuel variance and shrinkage?
How accurate is an automatic tank gauge?
How often should you reconcile fuel variance?
Where to start
If you only do one thing this month, compare your tank chart against a fresh calibration and check two dispensers against a certified measure. Those two checks settle most arguments about where the fuel went, and they cost far less than the fuel you are currently losing. If you want to see what continuous tank data and automated reconciliation look like in practice, the petrol and LPG automation product range covers the components, and our turnkey station automation work starts with a reconciliation review of the site as it stands today.
