Why reconcile card + sensor?

Card statements alone can’t confirm where or how much fuel actually entered a tank. Tank sensors tell you level changes; GPS confirms location and idling; trips explain consumption. Reconciling the three eliminates leakage and speeds audits. 1

Load your data

  • Fuel card CSV: date/time, station, liters, amount, card id, vehicle, driver.
  • Tank sensor: level readings per minute; calibration table.
  • Trips: start/end, distance, engine hours, idling.

Match to the right asset

  • Map card → vehicle → driver. Resolve duplicates and inactive cards. Require driver PIN + odometer on swipe to reduce errors.

Validate place & time

  • Check that the vehicle was within a fuel‑station geofence at swipe time (±10 min).
  • Flag off‑network or distance > 200 m from pump.
  • Cross‑check engine off during fueling if policy requires.

Confirm tank level change

Use sensor readings to compute Δlevel over the fueling interval. Compare against billed liters using the calibration curve and temperature correction if available.
Rule Pass Flag
Sensor increase ≈ billed liters (±7%) Within tolerance Outside tolerance
Only one fueling within 30 min Yes Multiple events
Vehicle stationary in geofence Yes Moving / outside

Reconcile against trip usage

Estimate expected consumption between fill‑ups from distance, load, and idling. Use class averages to sanity‑check outliers.

Open exceptions

Create an exception ticket for every mismatch: missing sensor increase, location mismatch, duplicate swipe, unusual liters, or fueling while moving.

Recover, report, improve

  • Recover funds for proven leakage; train repeat offenders.
  • Refine tolerances by vehicle class and station brand.
  • Publish weekly reconciliation KPIs.

KPIs

  • Matched swipes % (target: > 95%)
  • Leakage rate (liters & % of spend)
  • Average variance per swipe
  • Exceptions resolved in < 7 days

💬 FAQs

Use card + trip usage checks only; prioritize sensors for high‑spend vehicles.
±7% vs sensor is a practical starting point; tighten after 4 weeks by vehicle class.
Use calibration curves and ignore short spikes while the vehicle is moving; sample the median level during the fueling plateau.
Yes—stream card data via API, run rules nightly, and auto‑open exceptions with links to map and chart evidence.
Track separate tank sensors and require job tickets/images at issue time; reconcile like a station.
Tip: ask your fuel provider for station geo coordinates and pump identifiers. Matching improves dramatically.