A van carrying stock is a warehouse that moves, staffed by one person, unsupervised, with the door open several times an hour. Treat it as anything less and inventory accuracy will decay until nobody trusts the numbers.
The principle: the vehicle is a stock location
Every movement in and out must be recorded against the vehicle:
| Movement | Direction |
|---|---|
| Load-out from depot | In |
| Sale or delivery to customer | Out |
| Customer return | In |
| Damage and write-off | Out |
| Sample or free-of-charge | Out |
| Transfer to another vehicle | Out / In |
| Return to depot | Out |
| Cycle count adjustment | Either |
Any movement that happens without a record becomes variance. Most van inventory problems are not theft; they are unrecorded legitimate movements — a sample given away, a damaged case binned, a transfer between drivers at a layby.
If recording a damage takes six taps and a supervisor call, drivers will not record it. Every unrecorded movement becomes shrinkage on paper.
Load-out discipline
Load-out is where accuracy is won or lost.
- Scan or confirm at load, not after. A load recorded from the pick list rather than from what physically went on the vehicle inherits every picking error.
- Load in reverse route sequence so the first stop is at the door. It saves service time and reduces the rummaging that creates damage.
- Driver confirms the load. A driver who accepts a load without checking has no defensible position at settlement, and both parties know it.
- Segregate returns and empties physically, so they do not get counted as saleable stock.
- Handle late additions explicitly. Product added after the recorded load-out is the single most common source of settlement mismatch.
Counting without stopping the operation
Full physical counts of every van are disruptive and, done annually, useless for control. Use cycle counting.
Rolling partial counts. Each van counts a defined subset each week — highest value, fastest moving, or historically problematic SKUs. Every SKU is counted several times a year without any full stop.
Trigger-based counts. Count a SKU when variance exceeds a threshold, when a driver changes, after a promotion ends, or when a return pattern looks anomalous.
End-of-day blind counts for a small set of high-value lines. Blind — the driver counts without seeing the expected number — otherwise the count confirms the system rather than checking it.
Full counts at driver handover, vehicle change and period end.
Track count accuracy as a metric in its own right: how often the count matches, and by how much when it does not.
Understanding variance
Not all variance is theft, and treating it as such destroys the reporting you need.
| Cause | Signature | Fix |
|---|---|---|
| Picking error at depot | Consistent by SKU, appears at load | Scan verification at pick |
| Unrecorded damage | Sporadic, concentrated in fragile lines | Easier damage capture |
| Unrecorded samples | Small quantities, regular pattern | Sample workflow with allowance |
| Return not processed | Stock appears missing then reappears | Returns process at settlement |
| Unit-of-measure error | Large, exact multiples (cases vs units) | Configuration and training |
| Wrong SKU delivered | Two SKUs offsetting | Scanning at the stop |
| Theft | Persistent, high value, one route | Investigation |
Categorising variance before reacting to it is the difference between a control process and a witch hunt. Most operations find that the majority of early variance is process, not people.
Targets worth setting
- Inventory accuracy by count — above 98% of counted lines matching is achievable in a disciplined van operation.
- Value variance per route per week — set a threshold that triggers investigation rather than chasing every case.
- Unrecorded movement rate — variance that cannot be attributed to any category. Trending this down is the real programme goal.
- Time from variance to explanation — under 48 hours, or nobody remembers.
Technology that helps
Barcode scanning at load-out and at the stop. The single biggest accuracy improvement available, and the most resisted because it adds seconds per case. Introduce it on high-value lines first.
Offline-capable mobile inventory. Signal will fail. Stock movements must record locally and sync later without duplication.
Photo capture for damages and returns. Removes most disputes and takes one tap.
Automatic reconciliation at settlement. Expected versus actual computed by the system, with variances presented for explanation rather than discovered later in a report.
RFID or weight-based systems in specific sectors — gas cylinders, kegs, high-value equipment — where per-item tracking justifies the cost.
The organisational part
Van inventory accuracy is a management practice, not a software feature. It needs:
- A named owner of inventory accuracy, not shared between depot and finance
- Weekly variance review with the drivers concerned, in a non-punitive framing for process causes
- A clear, written policy on samples, damages and write-offs, with authority limits
- Consequences that are consistent and known in advance
- Recognition when accuracy improves, not only attention when it fails
Operations that get this right typically see variance fall sharply in the first quarter after introducing measurement, before people even change behaviour — because the act of counting reveals process gaps that were always there.
Frequently asked questions
How often should van stock be counted?
Cycle count weekly on a rotating subset, with full counts at driver changeover and period end. Annual full counts alone provide no operational control and arrive far too late to explain anything.
Should drivers be held financially responsible for variance?
Policies vary by jurisdiction and are often legally constrained; many are unenforceable. Regardless of legality, deduction-based approaches tend to drive concealment rather than accuracy. Clear process, visible measurement and consistent consequences work better.
What causes most van inventory loss?
In most operations, unrecorded legitimate movements — damages, samples, unprocessed returns and picking errors — outweigh theft substantially. Fix the recording process before assuming dishonesty.
Do we need barcode scanning?
It is the largest single accuracy lever, particularly at load-out. If full scanning is impractical, apply it to the highest-value and highest-variance SKUs, which typically covers most of the value at a fraction of the effort.
How do we handle promotional and free-of-charge stock?
Give it its own movement type with an authorisation limit and reporting. Free stock recorded as a sale, or not recorded at all, corrupts both inventory accuracy and margin analysis.
