BAMML Executive Insights

Shrinkage Is Not a Stock Problem, It Is a Supervision Problem

Industry: Fuel Retail / Forecourt · Primary pillar: Operational Performance · Secondary: Workforce Governance; Risk & Compliance

5 min read

Every forecourt carries losses. Wet stock variance, shop shrinkage, unexplained differences that surface at month-end and get written off because the alternative — finding out — takes longer than anyone has.

Written off often enough, a loss stops being an exception and becomes a budget line. At that point the organisation has quietly decided the loss is unavoidable.

It usually is not. It is just unattributed.

What is actually happening

Losses are recorded against a period. The month lost this much. Sometimes against a site, in a multi-site group.

Forecourt shop staff member doing a stock count at the shelves with a tablet

But no loss actually happens to a month. It happens at a moment — on a shift, at a pump or a till, under a named supervisor, with a specific set of people on duty. That is the level at which it could be explained, and it is almost never the level at which it is measured.

There is a second complication that makes the first one worse. Wet stock variance has entirely legitimate causes. Temperature affects volume. Meter calibration drifts. Deliveries vary within tolerance. A genuine reconciliation has to separate these from the illegitimate causes before it means anything.

Most sites do not make that separation. So every variance is treated as ambiguous — which, in practice, means every variance is tolerated.

Why it matters

An unattributed loss cannot be managed, and it cannot be acted on fairly.

Supervisor and attendant counting a till drawer together at the end of a shift, with a reconciliation sheet on the desk

That second point matters more than operators usually expect. Suppose you become confident that losses are not accidental. Acting on that requires evidence tied to individuals, times and events. A monthly site-level variance does not support a disciplinary process. It supports a suspicion — and a suspicion acted on without evidence creates a new problem, this time a procedural one, with its own cost.

So the operator is caught. The loss is real, the cause is probably known informally, and there is nothing on which to act. Meanwhile the people who are not responsible are working under the same cloud as the people who are.

Attribution solves both halves. It makes the loss manageable, and it makes any consequence defensible.

The Integrated View

Discipline What it contributes What is missing without it
Operational Performance Reconciliation at shift level, with legitimate variance separated out A monthly number nobody can act on
Workforce Governance Named accountability per shift, and a fair disciplinary process where needed Suspicion without evidence, applied to everyone
Financial Governance The loss quantified against the segment it erodes A write-off absorbed into a blended margin
Risk & Compliance Evidence that would support a consequence Exposure whichever way you act
Executive Decision Intelligence Variance reported by shift and supervisor, trended A budget line called shrinkage

What good looks like

Reconcile at shift handover rather than month-end. The tighter the interval, the smaller the population of possible explanations — and the more likely someone remembers what happened.

Supervisor and attendant going through a shift handover on a tablet at the forecourt pumps at dusk

Separate legitimate variance explicitly. Record temperature, calibration and delivery tolerances as known causes so that what remains is genuinely unexplained rather than merely unexamined.

Attribute what remains to a shift and a supervisor. Not as an accusation — as a record. Most of the value appears here, and it appears quickly, because attribution changes behaviour before anyone is ever disciplined.

Then report variance by supervisor, trended. If the pattern concentrates, you have something specific to address. If it is evenly distributed, you likely have a calibration or process issue rather than a conduct issue — which is also a useful finding, and a cheaper one.

One BAMML client carried stock and cash control weaknesses alongside margin pressure and VAT exposure. Tightening those controls formed part of a review that moved gross margin from 13.05% to 15.44% and operating expenses down 22%. Control discipline was not a side benefit of that result. It was a component of it.

What to do on Monday

  1. Ask when reconciliation actually happens. If the answer is month-end, that is the finding.
  2. Ask what the legitimate variance tolerance is, and whether it is documented. If nobody can state it, every variance is being treated as normal.
  3. Ask for last month's variance broken down by shift and supervisor. If that report cannot be produced, you have no basis for action of any kind.
  4. Check when meters were last calibrated, and whether that record is current.
  5. Ask what happens when a variance exceeds tolerance — is there a defined step, with an owner, or does it go into the write-off?

The objective is not to catch someone. It is to reach a position where, if there is something to catch, you are able to act on it properly — and where, if there is not, you stop paying for a problem you do not have.

Share this insight LinkedIn Email Book a conversation

← All Executive Insights