Fuel is the single largest controllable line item in European long-haul trucking — and also the easiest to steal. Mid-sized fleets can lose up to 22% of fuel spend to fraud (Aon Eurasia Risk Report 2024, KPMG European Fleet Survey 2025). Worse: those stolen litres are invisible in manual reports — average consumption looks “normal”.

In this article we’ll define ghost-pump fraud, walk through the working mechanics across Europe, and show how OBD2 tank-level matching against card transactions catches it within minutes.

What is a ghost pump?

Picture a standard refuel:

Driver hands over the fuel card.
Attendant prints a receipt for 60 L.
The tank actually takes 30 L.
The missing 30 L is siphoned to a side tank or grey market — split between attendant and driver.
Card transaction processes, accounting books it, trip closes.

The monthly report shows: 60 L consumption, within normal band. Because some other lane that month consumed below average; the deviation dissolves into the mean.

Industry case: A 100-truck fleet loses on average ~533 L per truck per year to fake refuels. At the €0.63/L EU pump average, that’s roughly €33,500/year of raw loss — from this single vector. Most CFOs never see this number because “fuel budget on target” is tracked as total spend deviation, not lane-level line items.

Why manual detection fails

Three structural reasons:

1. Missing granularity. Receipts aren’t trip-level records. Raw CSV from UTA, DKV, AS24 etc.: card number, amount, litres, station, timestamp. Which trip? From which tank level to which? Not captured.

2. The averaging illusion. An 18-tonne diesel doing 100 L/100 km shows a 5% deviation as 5 L/100 km. That deviation typically hides behind three plausible reasons: steep gradient, headwind + dorse load, ambient temperature. Fake refuels nest behind these three.

3. Customer resistance. Manual auditing — placing a controller at the pump or running periodic audits — is costly and brittle. About 30% of field auditors rotate to another fleet within 6 months; you never build a durable audit layer.

OBD2 + card-transaction matching

The right solution moves the raw signal off the field and onto the data.

Lognari continuously reads each truck’s OBD2 tank-level sensor. Standard refuel signatures:

Empty → half: 50–80 L rise
Half → full: 100–150 L rise
Topping up: 10–30 L

In parallel, fuel-card transactions arrive real-time or nightly sync via API: card number, litres, station GPS, timestamp.

The two signals are matched within the same minute:

Event (card)	Event (OBD2 tank)	Δ	Verdict
60.0 L · TR-Şile · 14:22	+58.8 L · 14:23	+1.2 L	OK
50.0 L · BG-Kapıkule · 09:14	+49.3 L · 09:16	+0.7 L	OK
60.0 L · RO-Kalvarya · 23:48	+28.2 L · 23:51	−31.8 L	🚩 FLAG

If Δ > 8%, an instant alert fires: which truck, which driver, which station. The alert is actionable.

How alerts become action: a 3-stage process

Stage 1 — Single event (drift tolerance): One deviation can be sensor error, hose blockage, or tank-shape variance. A single event does not trigger discipline; only a record is logged.

Stage 2 — Pattern (3 consecutive events): Same driver-station combo, 3 consecutive + consistent deviations → second-stage verification triggers. The driver sees it in the mobile app with reason-entry required. The station gets flagged on the analyst panel.

Stage 3 — Supplier decision: If the same station shows the pattern across 3+ different drivers, it enters supplier removal workflow. Step one: escalate via card provider (UTA, DKV, …) plus price renegotiation. If renegotiation fails, route to an alternative station on the corridor.

A real pilot — 95-truck EU export fleet

Over a 6-month Lognari pilot, three stations on the Romanian crossing showed a persistent 18–22% deviation pattern. The same 4 drivers had systematic Δ flagged at every refuel. The escalation path:

Month 1: Pattern detected, driver mobile app activated with reason-entry prompts. 1 driver stopped (evidence of behavioural change after warning).
Months 2–3: Other 3 drivers continued the pattern. Operations centre intervened — 1 driver exited by mutual agreement.
Months 4–5: Station negotiation → price discount refused → corridor rerouted to an alternative station 28 km off the original line.
End of month 6: Cumulative prevention €14,200; annualised projection €28,400/year from this single vector.

OBD2 device cost was recovered in 32 days. The next 24 months are pure positive margin.

Contract + audit layer

Two clauses became standard in customer contracts after the pilot:

Clause 1 — Δ tolerance: “For every refuel where consumption deviation exceeds 5%, the fleet operator retains the right to demand a wholesale-approved audit.”

Clause 2 — Data sharing: “The fuel-card provider must deliver transaction data via API within 24 hours of the event.”

These two clauses filter out fraud-exposed suppliers at the contract stage.

The CO₂ angle

Ghost-pump fraud is not just financial — it’s double-billed. A station running fake refuels also corrupts CO₂ reporting: your fleet books emissions for fuel that wasn’t actually burned. For CSRD reporting (mandatory 2027), this is a structural risk: data accuracy won’t survive audit.

OBD2 + card matching is the only architecture that grounds GHG reports in actually burned litres. For CSRD Scope 1 verification, this is the standard.

What’s next

If your fleet is 50+ trucks and you haven’t reported a single fraud case in the last 12 months, fraud is almost certainly happening — just not getting caught. Aon and KPMG reports suggest a positive case in 95%+ of fleets.

First step is small: a 30-day Corridor Pack pilot (€3,500), 50 trucks, OBD2 devices included. You leave the pilot with concrete loss figures + concrete prevention method. If you walk away, no charge.

Reach out via the contact section — we reply within one business day.

Statistics referenced from Aon Eurasia Risk Report 2024, KPMG European Fleet Survey 2025 and Lognari pilot data (anonymised). €/L values reflect Q2 2026 EU pump average; actual numbers vary ±8% by card provider and region.

Ghost pump: how to catch fuel fraud within minutes