Investigate fuel variance before calling it a loss

A fuel receipt and a tank reading may measure different things

A purchase record shows 120 litres delivered. The vehicle’s tank sensor rises by only 105 litres. A dashboard labels the 15-litre difference as fuel loss and assigns it to the driver.

That conclusion is premature. The readings may have been taken under different conditions, at different times or from an incorrectly associated sensor. Fuel may have been consumed between observations. The supplier record itself may need checking. The investigation must establish a comparable measurement basis before deciding what the difference means.

A useful fuel-control process preserves the discrepancy and makes it actionable without converting uncertainty into an accusation. It should help the relevant reviewers determine whether the issue is measurement, recording, operations, supply or something else supported by evidence.

Define the reconciliation boundary

Identify the vehicle, tank or fuel system, purchase transaction, opening observation and closing observation. Record their units and timestamps. A transaction can include more than one tank or equipment unit, so a vehicle-level total may need further allocation before comparison.

The Frappe HR — Vehicle Log documentation describes vehicle records containing odometer, fuel and service information. These records can support reconciliation. They do not establish that a sensor is calibrated, that a supplier delivered the recorded quantity or that a discrepancy represents theft.

Separate purchased volume from consumed volume. For a defined period, an illustrative balance of 200 litres opening stock plus 100 litres purchased minus 120 litres closing stock implies 180 litres used or otherwise unaccounted for within that boundary. Further evidence is needed to classify the balance; the arithmetic alone does not explain it.

A consumption comparison also needs distance, operating conditions and relevant auxiliary use. Do not compare a vehicle with substantial stationary equipment operation against one doing only road travel without acknowledging the difference.

Illustrative delivery discrepancy: fifteen litres remains a question

Consider a fictional refuelling event. The purchase record states 120 litres. The initial tank reading is 80 litres and the later reading is 185 litres, giving an indicated increase of 105 litres. The receipt-to-increase difference is therefore 15 litres.

Suppose the reviewer learns that the closing observation was taken on a slope and a later controlled observation reads 197 litres. The latter is 117 litres above the recorded opening value, leaving a three-litre difference against the purchase record if the observations are otherwise comparable.

That does not prove that slope caused exactly 12 litres of error. The reviewer must establish timing, consumption, measurement conditions and the reliability of the opening observation. The later reading is new evidence, not permission to rewrite the first reading or declare the case solved.

Build an evidence-led investigation

The fleet reviewer confirms the physical identity and event sequence. The technical owner checks sensor configuration, measurement conditions and known limitations. Finance verifies the supplier document, payment reference and whether the transaction was duplicated or corrected.

Where employee input is relevant, request a factual account and supporting records through the organisation’s approved process. Do not present an unverified dashboard label as a finding against the person. Restrict access to the case and separate operational investigation from any authorised employment process.

Each investigation should state what would resolve the discrepancy. Examples include confirmation of a duplicated purchase entry, corrected tank association or a technically supported measurement explanation. A generic “reviewed” status leaves the next analyst unable to understand the conclusion.

If the evidence remains insufficient, retain an unresolved or qualified outcome. A forced choice between “normal” and “loss” encourages unsupported certainty and makes later analysis less reliable.

Compare simple logs, telematics rules and a reconciliation queue

A controlled fuel log can be effective where transactions are few and the supporting records are accessible. It should preserve the original entries and any corrections. Its limitation is the effort required to align multiple time series and identify duplicate transactions.

A telematics rule can highlight abrupt changes or differences worth investigating. Its usefulness depends on the device, installation, sampling and configured context. A rule’s alert name should describe the observed discrepancy rather than claim a cause it cannot establish.

A reconciliation queue can combine fuel transactions, readings, journey information and reviewer decisions. The additional value is an organised path from alert to evidence and outcome. It should not simply display a larger list of alleged losses.

Use the same illustrative discrepancy to evaluate each approach. Ask whether a reviewer can identify the observation conditions and preserve an uncertain result. If the demonstration requires every variance to become a financial recovery amount, the workflow is overreaching.

Choose metrics that do not reward premature conclusions

Track the number and value or volume of discrepancies reviewed, the proportion resolved by evidence category and the age of unresolved cases. Define whether the denominator is transactions, refuelling events or vehicles. Those populations are not interchangeable.

For fuel efficiency, state the distance and consumption basis and exclude or separately explain incomplete periods. A missing closing reading should not produce a zero-consumption trip that artificially improves the average.

Review false alerts and their causes. Incorrect tank mappings, repeated imports and stale readings can create workload without improving control. Fixing those causes is a better result than training analysts to dismiss a noisy dashboard.

Do not report every resolved variance as a recovered saving. A correction to a duplicated entry improves the record but does not necessarily return cash or fuel. Financial effects require their own evidence and approval.

Test corrections and uncertain outcomes

Replay a duplicated supplier transaction and verify that it is linked to the original rather than counted as a second delivery. Then test a sensor reading associated with the wrong vehicle and confirm that correcting it triggers review of affected calculations.

Include an event with irrecoverably incomplete measurements. The system should retain the unknown result and the reviewer’s explanation. It should not fill missing readings with a convenient assumption to close the period.

If data arrives late, recalculate the proposed balance while preserving prior review versions. A case already used in a financial decision should not change silently. The responsible finance owner must assess any consequential correction.

For an assessment, use redacted transactions and sensor exports together with the current investigation procedure. The output should define comparison boundaries, evidence requirements and decision roles before new alert thresholds are introduced.

Fuel variance analysis is useful when it narrows an uncertainty into a supported explanation. The quality of that explanation matters more than the speed with which a dashboard assigns blame.

Next actions

1. Define the tank, period, units and observations used in each comparison.
2. Review a discrepancy for timing, identity and measurement-condition gaps.
3. Separate technical findings, transaction corrections and authorised financial decisions.
4. Test duplicate, corrected and unresolved cases before using the results in management reporting.