Short answer
Neither one catches fuel fraud alone. A fuel card records what the pump billed: litres, riyals, station and time. A tank sensor records what actually entered the tank and when it left. Fraud lives in the gap between the two, so the strongest control is a matching table that compares every receipt with the level rise the sensor saw.
Key takeaways
- A fuel card proves money was spent at a station; it cannot prove the fuel went into the vehicle it was meant for.
- A tank sensor proves how much fuel entered or left the tank; it cannot prove who paid, how much was paid, or why a drop happened.
- The control that catches fraud is a matching table: receipt litres next to the sensor’s level rise for the same fill, with a tolerance the fleet sets in advance.
- Fuel dispensers fall under accuracy class 0.5 in OIML R 117-1, with a 0.5% maximum permissible error, so a large gap is rarely a pump error.
- At Aramco’s October 2026 diesel price of 1.79 SAR per litre, a 45 L gap on one fill is 80.55 SAR, and repeated gaps add up quickly.
01What a fuel card proves, and where it stops
A fuel card, or a driver’s paper receipt, tells you what happened at the pump. You see the litres the dispenser counted, the amount in riyals, the station and the time. That is useful evidence for finance, and it closes the oldest gap of all: cash handed to a driver with no record. A card also lets the fleet put limits on spending, so a card used at the wrong hour or for an unusual amount stands out on the statement. But the card stops at the nozzle. It cannot tell you whether the 300 litres went into the truck on the card, into a second vehicle, or into a drum in the back. It cannot see diesel that is siphoned out an hour after a clean fill. And when a pump attendant and a driver agree to bill more litres than were dispensed, the card records the bill exactly as it was rung up. The card is honest about the payment, not about the tank.
02What a tank sensor proves, and where it stops
A fuel level sensor sits inside the tank and reports the level many times an hour through the tracker or site device. Software turns that level into litres using the tank’s shape and size, then marks a refuel when the level rises and a drop when it falls. So the sensor answers the question the card cannot: how much fuel actually arrived, and did any leave when the engine was off or the vehicle was parked? That is why it catches siphoning, a drum filled from the vehicle tank, and a fill that never reached the vehicle. The sensor has its own blind spots. It knows nothing about money, so it cannot tell you the price, the station, or whether a receipt is fake. It sees a drop but not its cause, so the honest word is an unexplained drop, not theft. And its reading depends on correct calibration, steady readings and a level vehicle, so small gaps need care before anyone is blamed.
03The matching table: receipt litres against the sensor’s rise
The method is simple enough to run in a spreadsheet. For every fill, write one row: vehicle, date and time, litres on the receipt, and the level rise the sensor recorded for the same fill. The gap is receipt litres minus sensor litres, and the gap percentage is the gap divided by the receipt litres. Then compare it with a tolerance you agree before the first review, not after a dispute. The pump side is tight: OIML R 117-1 puts fuel dispensers for motor vehicles in accuracy class 0.5, with a 0.5% maximum permissible error for the complete measuring system. Most of your tolerance is therefore for the sensor, the calibration table, fuel temperature and a vehicle that is not level. Many fleets start with a band of a few percent and tighten it once the readings prove steady. Two more rows matter: a receipt with no matching rise at all, and a drop with no receipt, which only the sensor can show.
04A worked example in riyals
Take one diesel truck over one week, with a tolerance of 5% agreed in advance. This is an illustration, not customer data. Fill one: the receipt says 220 L and the sensor saw 216 L, a gap of 4 L or 1.8%, inside tolerance. Fill two: the receipt says 300 L and the sensor saw 255 L, a gap of 45 L or 15%, which needs a question. Fill three: 180 L against 177 L, a 3 L gap, fine. Fill four: the card shows 250 L, but the sensor shows no rise at all, so the fuel went somewhere else. At Aramco’s October 2026 diesel price of 1.79 SAR per litre, fill two is 80.55 SAR and fill four is 447.50 SAR. That is 295 questioned litres, or 528.05 SAR, from one truck in one week. Finally, the sensor shows a 60 L drop on Thursday night while the truck was parked, with no receipt near it. The card statement would never show that row.
05Choosing: card, sensor, or both
If you have neither today, start with whichever closes your biggest gap. A fleet that still pays drivers cash for fuel should fix the payment record first, because without a record of what was bought there is nothing to match. A fleet with a card already in place but unexplained consumption should add sensors to the vehicles that burn the most fuel or park overnight in open yards, since those are where siphoning and second-vehicle fills hide. Sites that store diesel in their own tanks need a sensor on the site tank too, because a depot tank that feeds vehicles is the other end of the same match: the litres that leave the depot tank should appear in a vehicle tank. In the end, both records together are what make a review fair. The driver who fills correctly is cleared by the numbers, and a real gap is shown with a date, a time and an amount instead of a suspicion.
06How this works on ENTEK IoT today
ENTEK IoT holds both sides of the match, and a person does the matching. In Fleet, drivers submit fuel receipts from the driver portal: the photo is read automatically by AI, checked with a simple math cross-check, and then an admin approves or rejects it, a decision that is locked once made. Each receipt shows litres, SAR, station and location. On a vehicle that reports fuel data, the vehicle detail screen shows fuel history and refuel and drop events. For site and depot tanks, Fuel Tank shows refuel and drop events with litres before and after, marks a tank Low below 35% and Critical below 20%, and can raise a sudden-drop alert. Rules are checked every 30 seconds and alerts reach named people by email, SMS or WhatsApp, usually within about a minute. There is no automatic receipt-to-sensor match and no Fleet report export today, so the matching table is kept in your own sheet. Sensors and trackers can be sourced through the ENTEK marketplace and RFQ at entek.ai.
| Question | Fuel card or receipt | Tank sensor |
|---|---|---|
| How much was paid, and where? | Yes: litres billed, SAR, station, time | No |
| Did the fuel reach this vehicle’s tank? | No | Yes: the level rise for that fill |
| Fuel put into a drum or a second vehicle | Not visible | Visible as a receipt with no matching rise |
| Fuel siphoned while parked | Not visible | Visible as an unexplained drop |
| Pump billed more than it dispensed | Records the bill as rung up | Visible as a gap between receipt and rise |
| Is the receipt itself genuine? | Partly: the statement shows the station and time | No |
| Why the fuel left the tank | No | No: it shows the drop, not the cause |
| Fill | Receipt litres | Sensor rise | Gap | Gap % | Gap value | Verdict |
|---|---|---|---|---|---|---|
| 1 | 220 L | 216 L | 4 L | 1.8% | 7.16 SAR | Within tolerance |
| 2 | 300 L | 255 L | 45 L | 15% | 80.55 SAR | Ask the driver and the station |
| 3 | 180 L | 177 L | 3 L | 1.7% | 5.37 SAR | Within tolerance |
| 4 | 250 L | 0 L | 250 L | 100% | 447.50 SAR | Fuel did not reach this tank |
| Thursday night | No receipt | 60 L drop while parked | Not applicable | Not applicable | 107.40 SAR | Unexplained drop: check the site |
Receipt litres vs sensor rise, per fill
Worked example in this article (illustration, not customer data).
Figures and sources
- Aramco in-Kingdom retail diesel price for October 2026: 1.79 SAR per litre (Source: Aramco, In-Kingdom retail fuels, as of 2026-10-19)
- Aramco in-Kingdom retail price of Gasoline 91 for October 2026: 2.18 SAR per litre (Source: Aramco, In-Kingdom retail fuels, as of 2026-10-19)
- Aramco in-Kingdom retail price of Gasoline 95 for October 2026: 2.33 SAR per litre (Source: Aramco, In-Kingdom retail fuels, as of 2026-10-19)
- Accuracy class of fuel dispensers for motor vehicles (other than LPG): Class 0.5 (Source: OIML R 117-1:2007, Dynamic measuring systems for liquids other than water, Table 1, as of 2026-10-19)
- Maximum permissible error for a complete class 0.5 measuring system (line A): 0.5% (Source: OIML R 117-1:2007, Dynamic measuring systems for liquids other than water, Table 2, as of 2026-10-19)
Questions
Is a fuel card enough to stop fuel theft?
No. A fuel card controls and records the payment, but it cannot see where the fuel went after the nozzle. Filling a second vehicle, filling a drum, or siphoning while parked all stay invisible on the card statement.
If I install fuel sensors, do I still need fuel cards or receipts?
Yes. The sensor shows how much fuel arrived but not how much you paid for. Without the receipt you cannot see that a station billed 300 L for a 255 L fill. You need both records to match.
What gap between the receipt and the sensor is normal?
Agree a tolerance before the first review. The pump’s own limit is small: OIML R 117-1 sets a 0.5% maximum permissible error for fuel dispensers. The rest of the band covers sensor calibration, fuel temperature and a vehicle that is not level. Many fleets start with a few percent and tighten it as readings prove steady.
Does ENTEK match receipts to sensor readings automatically?
Not today. ENTEK IoT holds both sides: AI-read driver receipts that an admin approves in Fleet, and refuel and drop events from the fuel sensor. A person puts the approved receipt next to the refuel event and keeps the matching table in their own sheet.
Can the sensor prove a driver stole fuel?
No. A sensor shows that fuel left the tank, when, and how much. It cannot show who took it or why. Treat a gap as a question to investigate with the driver, the station and the site, not as a verdict.
Sources
Related insights
- Inventory Reconciliation for Fuel Tanks: Finding the Loss That No Alert Fires On
- Refuelling or Theft? Why Flow Rate Alone Cannot Tell Them Apart in Tank Telemetry
- Fleet Fuel Numbers Do Not Add Up? When CAN Bus Is Enough and When You Need a Sensor
- How Fast a Legitimate Refuel Actually Is, and Why the Rate Still Cannot Judge It
- Fuel Sensor Showing the Wrong Litres? Datasheet Accuracy Versus the Field