Van inventory: why it never balances on Friday

At six on Friday evening the truck rolls back into the yard and the part nobody enjoys begins: reconciling. It left with 480 units, 412 were invoiced, 51 are still on board. Seventeen are missing. Nobody stole anything. Seventeen are still missing.

The scene repeats itself at food distributors, beverage wholesalers, parts suppliers and cleaning-product companies. The product changes; the ending does not. The supervisor deducts the difference, the rep argues, and Monday starts again with an opening count that is already wrong.

The truck is a warehouse, but nobody treats it like one

A warehouse has a door, an owner, working hours and a count somebody signs. A truck has all of that except the count. It opens and closes twenty or thirty times a day, on the street, in a hurry, and every opening is an inventory movement that almost never gets recorded at the moment it happens.

In United States retail, shrink — the gap between what the books say exists and what the count actually finds — was 1.6% of total sales in fiscal year 2022, about $112.1 billion, according to the National Retail Federation's National Retail Security Survey 2023 (accessed 3 September 2026).

That figure comes from stores, not trucks, and it is worth saying so. But it is a useful, uncomfortable reference point: if a stationary warehouse with cameras and one counter loses that share, a vehicle unloading at the curb has no reason to behave better. The surprise is not that 17 units are missing. It is that anyone expected otherwise.

A second number from the same survey changes the conversation: internal and external theft together account for nearly two-thirds (65%) of that shrink, according to the NRF release announcing the report. The remaining third is process. Paperwork, counts and transfers done badly. A third of what goes missing was never stolen by anyone.

The four leaks that break your Friday

When a discrepancy always shows up in the same place, it is not a mystery — it is a process without an owner. In a field-sales operation the leaks are usually four, and none of them has anything to do with how honest the rep is.

1. Returns

The customer hands product back during the visit: it arrived damaged, they over-ordered, they switched brands. The unit goes back on the truck, but the invoice is already issued and the credit note gets written at the office on Monday. For three days that unit exists twice: physically on board, sold in the system.

Reverse flow is not a rounding error. In US retail, consumer returns came to 15.8% of sales in 2025, roughly $849.9 billion, down from 16.9% in 2024, according to the NRF and Happy Returns release dated 15 October 2025.

Different channel, different kind of return — agreed. What it shows is scale: merchandise moving backwards is comparable in size to merchandise moving forwards. A system that only knows how to record a sale is blind to half of what moves.

2. Samples and trade support

The tasting, the courtesy unit, the pack left behind so the shopkeeper will try it. It leaves the truck, produces no invoice and rarely produces a record. By month end that is forty or fifty units the system still believes are sitting on board.

3. Expired or damaged stock

The case that split open under hard braking, the carton that sat in the sun, the lot that expired inside the truck because it had been riding around for three weeks. The rep sets it aside, sometimes drops it at the depot without telling anyone, and the write-off happens late or never.

4. Transfers between reps

This is the quietest leak of all. One rep runs out mid-morning, another has surplus, and ten cases change hands in a parking lot. Nobody signs anything. Both trucks are now wrong, and the discrepancy shows up split between two people who cannot explain it.

The problem is not counting — it is when you count

Almost every operation counts. The mistake is counting only twice, at load-out and at check-in, and expecting the middle to be reconstructed from memory twelve hours later. Thirty stops fit between those two counts, and a rep's memory is not an inventory system.

The working rule is simple: every movement gets recorded where it happens and when it happens. Not at end of day, not back at the office, not on Monday. If the return is captured at the customer's door, with a photo and a reason code, it stops being an argument and becomes a data point.

What to measure, and when

• At load-out: opening inventory by SKU, signed by the rep. No signature, no departure.
• At every stop: sold, returned, given as a sample, flagged as damaged. Four categories, not one.
• At transfer: who hands over, who receives, what and how much — with confirmation from the receiving side.
• At check-in: physical count against theoretical, with the gap stated in units and in money.
• Weekly: inventory accuracy per rep and per SKU, not just a fleet total.

That last line is the one that changes behaviour. A fleet-wide discrepancy points at nothing; the same metric broken out by rep and by product shows within two weeks whether the problem is one person, one fragile SKU or a route that loads too much.

What a week that balances actually looks like

It does not look like a week with no differences. It looks like a week in which every difference has a name: nine units returned at stop 14, three samples at stop 22, six cases transferred to the north-side rep. The total is still not zero, but it is no longer a hole.

A route inventory is not judged by how far it is out, but by how much of the gap is explained.

Where to start on Monday

Pick one truck and one week. Record the four categories at every stop, on paper if that is what you have. On Friday, compare that truck's discrepancy against the rest of the fleet. That contrast, not a procedures manual, is what convinces the team.

Everything else comes after: requiring confirmation on transfers, closing the day with a physical count, reviewing accuracy by SKU. But the order matters. First make the movement visible; then you can discuss accountability with data instead of suspicion.

Friday will still be reconciliation day. The difference is whether that hour is spent hunting for 17 missing units or reading a report that already says where they went.