Wetstock reconciliation checks that fuel movements balance: opening stock + deliveries − dispensed volume = closing stock. The gap between that expectation and the actual tank reading is your variance. Small alternating variance is measurement noise. Variance that is persistent, growing, or consistently one-directional has a physical cause — and there are only four realistic candidates.
1. What wetstock reconciliation is
“Wetstock” is simply the liquid fuel you own. Reconciliation is the act of proving that the amount you should have matches the amount you do have. It is stocktaking, with the complication that your stock is invisible, expands when it warms up, and is measured by two entirely different instruments at either end of the process.
Every station does this in some form. The difference between stations is how often, how accurately, and whether anyone acts on the result. A station reconciling monthly from handwritten sheets is technically performing reconciliation, but by the time a discrepancy appears, the thirty days in which it could have been investigated are gone.
2. The formula, and where people get it wrong
For each product, separately — never combined:
Expected closing stock = opening stock + deliveries received − volume dispensed
Variance = actual closing stock − expected closing stock
Variance is normally expressed as a percentage of throughput, not of tank capacity — because a 50-litre discrepancy against 20,000 litres sold is a rounding error, while the same 50 litres against 500 litres sold is an emergency.
Four mistakes account for most bad reconciliation:
- Using the delivery docket instead of the measured delivery. The docket is what the supplier says they delivered. Your pre- and post-drop tank readings are what actually arrived. If you reconcile using the docket figure, you have quietly assumed away one of the four things you are trying to detect.
- Combining products. Gasoline 95 and diesel each need their own reconciliation. Add them together and a positive error in one masks a negative error in the other, which is how a real leak hides for months.
- Using an inaccurate tank chart. Converting a level reading to a volume requires the correct strapping chart for that specific tank. A wrong chart produces a permanent phantom variance that no investigation will ever explain, because the fuel is fine and the arithmetic is not.
- Ignoring the water level. Water sits under the fuel and a level probe will happily include it in total volume unless it is measured separately. Rising water can look exactly like fuel appearing out of nowhere.
3. What tolerance is reasonable
There is no single legally universal figure, and any vendor quoting one as gospel should be treated with suspicion. In practice, well-run stations with functioning automatic tank gauging and calibrated meters typically operate with variance well under half a percent of throughput, and often around a tenth of a percent, once measured over a long enough period to average out noise.
The number matters less than two other things. First, the cumulative trend: daily variance is noisy and individually meaningless, but a 30-day rolling total that drifts steadily in one direction is signal. Second, your own baseline: once you know what normal looks like at your site, with your equipment, a departure from it tells you more than any industry benchmark.
This is also why comparing sites is so powerful. If three of your forecourts sit at 0.05% and the fourth sits at 0.3%, you do not need a theory about acceptable tolerance. You need to visit the fourth one.
4. Diagnosing the cause from the pattern
Once you have reliable daily figures, the shape of the variance narrows the cause quickly.
| Pattern | Most likely cause | First check |
|---|---|---|
| Alternating +/−, averaging near zero | Normal measurement noise | Nothing. Watch the rolling average. |
| Steady negative, one product, every day | Leak, or meter over-reading | Meter calibration first — it is cheaper to rule out. |
| Negative only on certain shifts | Staffing-related loss | Transaction and authorisation logs for those shifts. |
| Large negative on delivery days only | Short deliveries | Pre/post drop readings vs docket, per delivery. |
| Persistent positive | Meter under-reading, or wrong tank chart | Verify the strapping chart against tank spec. |
| Level falls with no sales recorded | Unauthorised dispensing, or loss from tank | Quiet-period gauge data and pump authorisation logs. |
| Variance appears suddenly on one date | An event — repair, recalibration, chart change | Maintenance and configuration change log. |
Note the ordering in that last column. Always rule out meter calibration before commissioning a tank test. A dispenser meter drifting 0.3% high produces exactly the signature of a slow leak, and calibration costs a fraction of what a tightness test and excavation do.
One important limitation: reconciliation gives you leak indication, not leak certification. If the evidence points at loss of containment, that becomes a job for a qualified contractor and a formal tightness test. Software finds the problem; it does not sign off on a tank.
5. The temperature problem
Fuel expands as it warms. A tanker delivering in the afternoon heat delivers warm fuel that occupies more volume than the same mass will once it has cooled in an underground tank overnight. Nothing has been stolen, but your reconciliation will record a loss.
This is why gauge probes measure product temperature, and why volumes can be corrected to a reference temperature before comparison. Without temperature data, a station in a hot climate generates a systematic seasonal variance that is entirely physical, and staff learn to dismiss variance reports as unreliable — which is far more damaging than the variance itself, because it teaches everyone to ignore the one alarm that matters.
6. Why daily beats monthly
Monthly reconciliation tells you that something went wrong during a thirty-day window. Daily reconciliation tells you which day. Given attendant rotas, delivery dates and maintenance records, knowing the day usually means knowing the cause.
Daily reconciliation is only practical when dispensed volume is captured electronically. Adding up handwritten pump readings every night is possible but it will not survive contact with a busy forecourt for more than a fortnight. This is the real argument for dispenser automation: not that it produces prettier reports, but that it makes daily reconciliation cheap enough to actually happen.
There is a second effect that owners consistently underestimate. When staff know reconciliation happens daily and attributes volume to a named attendant and shift, behaviour changes before any investigation takes place. A meaningful share of the improvement stations see in the first month is deterrence, not detection.
7. Where to start if you have nothing
You do not need to buy a system to begin. Starting this week, with a dip stick and a notebook:
- Reconcile one product only — whichever you most suspect. One product done properly beats all of them done badly.
- Dip at the same time every day, ideally before opening, with the pumps quiet.
- Dip immediately before and after every delivery and record what actually arrived, alongside the docket figure.
- Record pump meter totals at the same fixed time each day, per nozzle.
- Verify your tank chart against the tank’s specification before you trust a single volume figure.
- Keep a cumulative running total of variance, not just daily figures. The trend is the signal.
- Give it thirty days before drawing any conclusion. Two weeks of data will mislead you.
After a month you will know whether you have a real problem, and roughly what kind. That is a much better position from which to evaluate an automation quote — including ours — than a vague suspicion that something is wrong.
Related: Tank monitoring & wetstock · Dispenser automation · Readiness checklist
General guidance, not engineering or regulatory advice. Suspected loss of containment must be investigated by a qualified contractor under the requirements applicable in your jurisdiction.