Fuel loss is uncomfortable to investigate because the investigation itself carries a cost. Accuse the wrong person and you damage a crew you depend on; accuse nobody and the loss continues. The way out is to stop treating it as a question about people and start treating it as a question about data, because every pattern of loss leaves a different signature in the records. Once you can tell which signature you are looking at, you know whether you have a measurement problem, a process problem, or something that genuinely needs confronting — and you know it before anyone is named. You can work through what your own figures show in a free 30-minute session.
Diesel Loss: Find the Pattern Before You Look for a Person
Each type of fuel loss leaves a distinct signature in delivery, tank, and issue records. Reading the signature tells you what you are dealing with.
Five Patterns and How Each Shows Up
Described at the level you need to detect them. What matters here is the signature, not the method.
Signature: tank level after delivery consistently lower than the docket quantity implies, particularly with one supplier or one delivery window.
Detect by: dipping before and after every delivery and comparing the gain against the docket over time, not on a single occasion.
Signature: tank level falling faster than recorded issues explain, with the gap appearing between rather than during recorded activity.
Detect by: dipping on a fixed routine independent of issues, so tank level is known rather than inferred from paperwork.
Signature: one or more machines showing consumption per productive hour well above identical units doing comparable work.
Detect by: comparing like with like — same model, same site, same duty — rather than against a fleet average.
Signature: issues recorded against an asset whose meter reading did not move, or moved far less than the fuel implies.
Detect by: requiring a meter reading with every issue, which makes this immediately visible rather than invisible.
Signature: a stable unexplained percentage that everyone has quietly accepted as normal evaporation or measurement error.
Detect by: asking whether that percentage has ever been tested. Accepted variance is where persistent loss survives longest.
Why It Persists Even When People Suspect It
Nothing is reconciled
Without opening level, deliveries, issues, and closing level in one place, there is no figure to be alarmed by. Suspicion has nothing to attach to.
The cost of being wrong is high
Managers avoid raising it because an accusation that does not hold damages relationships with people the site depends on daily.
It is budgeted for without being named
Once a loss rate is built into the fuel forecast, it stops appearing as a variance and becomes an expectation.
Responsibility is split
Procurement buys it, the site stores it, operators use it, and finance pays for it. A loss that crosses four functions belongs to none of them.
Get the Numbers That Make This Answerable
Deliveries, dips, and issues recorded against machines with meter readings, reconciled monthly.
A Sequence That Does Not Start With an Accusation
One
Fix measurement first. Consistent dips and recorded issues with meter readings. Most apparent loss reduces at this stage, which tells you it was never theft.
Two
Establish the baseline. Run reconciliation for a period without acting on it, so you know what normal looks like before deciding what is abnormal.
Three
Narrow by pattern, not by person. Identify which signature the remaining variance matches and which site, shift, or stage it concentrates in.
Four
Change the process before confronting anyone. Tightening controls often ends the loss without a confrontation, which is usually the better outcome for the site.
Five
If it continues, escalate properly. By this point you have evidence rather than suspicion, and your organisation's HR and legal process should govern what happens next.
Frequently Asked Questions
How much of typical fuel variance is actually theft?
Less than most people assume before they measure properly. In fleets with no reconciliation, a large share of apparent loss turns out to be unrecorded issues and inconsistent dipping, which is why fixing measurement comes first. You can discuss how to run that first phase on a call with our team.
Should we announce that we are starting to track fuel?
Generally yes. Announcing it as a measurement improvement often reduces loss on its own, and it avoids the impression that people were being watched covertly, which damages trust even when nothing is found.
Is locking hardware the answer?
It helps at the tank, but it does nothing about fuel over-recorded into a machine or booked against equipment that was not running. Physical controls and record controls address different patterns.
What if the variance points at a supervisor?
Then it is a matter for your HR and legal process rather than for a maintenance system. The value of the data is that it gives an evidence base, not that it determines the response.
Where does the first improvement usually come from?
Recording every issue against a specific machine with its meter reading. It closes the largest gap and makes several of the patterns above visible immediately. Start with a free trial.
Measure First, Narrow Second, Confront Last
Fix dipping and issue recording, establish what normal variance looks like, identify which pattern the remainder matches, tighten the process — and only escalate to people once you are working from evidence rather than suspicion.
.png)