A fuel efficiency comparison between drivers is built by dividing the model’s expected fuel efficiency by the actual fuel efficiency each driver achieved, not by comparing liters against liters. That ratio is the only thing that lets you put someone driving a light utility vehicle and someone driving a one-ton pickup in the same table. The analytical work is done by the fleet manager: the Fuel module provides the fuelings, the liters, the meter’s kilometers and the expected fuel efficiency of each model; building the comparison and reading it are human decisions.
Comparing the absolute consumption of two drivers is the fastest way to reach the wrong conclusion. Two people who fueled 1,200 and 1,900 liters in the same month are not comparable if one drove 4,000 km of steady highway and the other 3,100 km of urban delivery with two hundred stops. The higher number does not point to the worst driver: it points to the most demanding operation.
The same mistake shows up in a more sophisticated version: comparing liters per 100 km without looking at which vehicle each person drives. A model with a larger engine consumes more by design, so if the driver of the heaviest unit comes in last, the table is not measuring driving: it is measuring fleet assignment.
Expected fuel efficiency is master data by model: how many liters per 100 km that unit should consume under normal conditions. It lives in the Models Master and is the same value the module uses as the third fueling control, when it compares actual consumption against expected consumption and flags the fueling for REVIEW if the deviation is significant.
That data point serves two functions that should not be confused. As a control, it operates fueling by fueling and detects one-off anomalies. As an analytical reference, it normalizes: it converts the consumption of different vehicles into a common scale. If it is entered incorrectly, both functions fail at once without warning. A value copied from the manufacturer’s spec sheet, measured in a lab, leaves the whole fleet below expectations and makes the comparison useless. Before moving on, review how expected fuel efficiency is defined.
The index is a ratio between expectation and reality, on a base of 100:
Index = (model’s expected fuel efficiency ÷ driver’s actual fuel efficiency) × 100
Since fuel efficiency is expressed in liters per 100 km, a lower number is better. That is why the expected value goes in the numerator: that way the index reads in the intuitive direction. An index of 100 means the driver consumes exactly what was expected for their vehicle; above 100, they consume less; below, more.
Actual fuel efficiency is calculated with the liters fueled in the period and the meter’s kilometers between the first and last fueling in the window. Two criteria change the result and should be fixed before calculating:
Exclude fuelings under REVIEW until they are resolved. A fueling flagged by geolocation, tank capacity or someone else’s card contaminates the calculation in either direction. If you do not know whether those liters went into that vehicle’s tank, do not use them to evaluate anyone.
Close the period by fueling, not by calendar. If you cut off on the 30th with a full tank, you are assigning liters to kilometers that will be driven the following month.
With small volumes, one month is not enough: a driver with four fuelings has an index that moves several points with a single poorly recorded meter reading. A rolling three months is the minimum for the number to stop being noise.
Route type. The biggest distortion. Urban delivery with frequent stops consumes considerably more than steady highway driving, and no normalization by model corrects for it. If your fleet mixes profiles, the valid comparison is within each one; Base or cost center usually approximate it well.
Load carried. A fully loaded vehicle consumes more than the same vehicle empty. In distribution it varies by day and is rarely recorded precisely enough to correct for: the honest approach is to acknowledge it as a margin of uncertainty.
Seasonality. Consumption rises in summer due to air conditioning and in winter due to cold starts. It affects the whole fleet at once, so it does not change the order within a month, but it breaks any comparison between different months.
Mechanical condition. The one that creates the most unfairness, because it gives the driver a number that belongs to the unit. A clogged filter, low tire pressure or a dragging brake push the index down without the person doing anything differently.
When an index is consistently low there are three hypotheses and only one is about the person. The order in which you rule them out determines whether the conversation that follows is fair.
The data is wrong. A poorly recorded meter reading, missing fuelings, unresolved fuelings under REVIEW, outdated expected fuel efficiency. It is ruled out by reviewing records, not by interpreting them.
The vehicle has a problem. It is ruled out by checking whether the low index follows the vehicle or the person. If another driver drove that unit and also came in below, the case belongs in a Corrective ticket in the ticketing system, not in an HR conversation.
Driving explains the difference. You only get here with the previous two ruled out. And you still do not know which part: it may be sustained speed, harsh acceleration, or engine idling while waiting for loading and unloading, which is the most common cause and the one the driver does not perceive as consumption.
The conversation starts by showing the data and asking, not by drawing conclusions. The manager arrives with the period’s index, the model’s expected fuel efficiency and the list of fuelings; the driver arrives with the context the system does not have.
Show the benchmark, not the ranking. Telling someone they are tenth out of twelve invites them to argue with the table. Telling them their vehicle consumes 12% above what is expected for that model invites them to explain why.
Separate the finding from the consequence. The first conversation is diagnostic. If there are formal consequences tied to consumption, they must be written into the fuel policy and signed in advance.
Record what the driver contributes. Half the time the explanation is operational and corrects your own analysis: a route that changed, an added load, a unit that has been acting up. That last case is a ticket, not a warning.
The order matters more than the technique. Comparisons that end in conflict did not fail because of the formula: they failed because the table was shown before the data could support it.
First, the expected fuel efficiency of each model. Without that reference reviewed there is no comparison, only a list of liters. It is the step that takes longest and the one most people try to skip.
Second, the quality of the kilometer records. The meter is the denominator of the entire calculation. If it is filled in by eye at month-end, the index measures administrative tidiness, not driving.
Third, fuelings under REVIEW resolved. A fleet with open invalid fuelings cannot evaluate drivers: it does not yet know which fuel went into which vehicle. Close that front using the approach in how invalid fuelings are managed.
Fourth, the calculation and its internal reading. Calculate, review the three hypotheses and correct. This first period is not shared with anyone: it serves to uncover the errors in your own method at zero cost.
Fifth, only then the conversation. And start with the cases where you have already ruled out the data and the vehicle. Credibility is decided in the first three: if in any of them the driver shows that the problem was mechanical, the comparison loses authority for months.
