Customer returns recorded on the dispatch grid: the ROI of a number that finishes its journey

Type the return into the very cell that was delivered, so every kilo a supermarket sends back is received at cost the moment the delivery note is submitted: how far the stock ledger drifts without it, how to measure the two labour streams, and why the largest effect — books that are right — is left out of the total.

The piece on the dispatch grid ended where the goods leave the warehouse: delivery notes, pick lists, trucks. This one follows the other half of the trip — the truck coming back, carrying the crates the supermarkets sent back.

At a fresh-fruit distributor selling to supermarket chains, returns happen every day: bruised fruit, fruit near its date, the wrong grade. They are not large — but they come back to the warehouse by a road the books cannot see. This piece works out the ROI of recording returns right on the dispatch grid, so that every kilo that comes back gets a receipt and a price on the day it arrives.

1. Where returns go today, and why the books do not see them

Before the change, the dispatch grid — the delivery-note × item matrix inside the ERP — recorded only what was DELIVERED. The driver typed what the supermarket returned into a "Returned" cell on the phone delivery screen, but that number was used only to reconcile with the customer. The goods came back for real; the stock ledger never heard of them.

Measured over the last 60 days, at a business running about a hundred delivery notes a day:

• 2.2% of delivery lines carry a return, and 5.9% of delivery notes have at least one returned line;
• 13% of customers returned something at least once;
• returned quantity is about 0.47% of delivered quantity;
• and before the change, 100% of it stayed out of the stock ledger.

0.47% sounds like a rounding error. It does not disappear — it does three things:

1. Book stock falls a little further below real stock every day. At the next count the warehouse is "somehow" over, and nobody can say why. The counter books the surplus to make the numbers agree, and where it came from is lost.
2. Cost of goods sold in the period is booked too high. Goods that came back still count as sold, then return to the books in a later period as a stock-count difference — the right total, at the wrong time, under the wrong label.
3. Two people keep two copies of one number. The driver keeps it on the phone, the dispatcher looks at the grid, and the stock accountant adds them up by hand when needed.

2. What the change is

The return is typed where people were already looking: the delivery cell on the grid.

1. A negative number in the cell = what the customer returned. A cell delivering 10 with 2 returned shows "10 (2)" in bold red; the delivered figure stays as it was. No new column, no new screen.
2. Two screens, one number. A return typed on the grid fills the "Returned" cell on the phone delivery screen, and the other way round — the driver and the dispatcher no longer keep two copies.
3. Submitting the delivery note records the return. The system creates a receipt into a salvage warehouse at the current cost — debit inventory, credit cost of goods sold, exactly like a sales return. Cancel the delivery note and the receipt is cancelled with it.

None of it asks anyone to learn another screen. It is the same principle as the dispatch-grid piece: a change has to live inside a motion people already make, or it gets skipped.

Two questions follow straight away — does stock in the salvage warehouse still count as stock you can sell, and where does the stock count find it — and each has a piece of its own: sellable stock versus stock on hand and the daily stock count screen.

3. How far off the books are — the largest effect, and not money

With the default assumptions in the calculator above — 100 delivery notes a day, an average cost of 1.5 million VND per note, 30 delivery days a month — returns carry a cost of about 21.2 million VND a month. Before the change, all of it sat in the warehouse without the books knowing.

The chart in the calculator draws one month of it. The sawtooth is the gap between real and book stock: it grows a little every delivery day and drops to zero at each count — with a count every five days, it peaks at about 3.5 million. The staircase is that gap being booked as surplus of unknown origin at each count, adding up to 21.2 million by the end of the month. At a gross margin of 25% on cost, that is about 1.9% of the month's gross profit booked as cost of goods by mistake.

It is tempting to add this to the benefits. The calculator does not. The goods were not lost, and the year-end total still agrees; what was wrong is timing and label. But the label is worth more than it looks: from now on every salvage receipt is born from a delivery note, so the cost of returns can be traced to each of the 13% of customers who return goods. Knowing who returns most is an input to a commercial conversation — but it is not a cash stream, and this piece does not count it.

4. First stream: less reconciling each returned line by hand

How to measure it. Time one end-of-day session of whoever reconciles, once before switching on and once after.

5. Second stream: less dealing with unexplained surplus at the count

How to measure it. At the two counts before switching on, count the surplus lines the counter could not explain. After switching on, that number should be close to zero for items with returns.

When hours become money. As the piece on keyboard entry put it, freed hours become money only when they go into work that has a price. Otherwise read both streams as capacity, and do not count them a second time somewhere else.

6. Adding it up

In money this is a small improvement, and the numbers say so. At 8 million VND to make the change — a few days' work on a system that already has the dispatch grid and the delivery screen — it pays back in about month 7.4. The reason to do it is not in the table: it is in section 3, in a stock ledger and a gross-profit report that are right day by day from now on.

7. What this ROI stands on

Just two numbers of minutes. Halve both — 1.5 minutes per returned line, 1 minute per surplus line — and 0.54 million VND a month is left, payback in about month 14.8. Set both return ratios to zero and every stream, and every line on the chart, goes to zero: if customers return nothing, recording returns is worth nothing. A model that cannot collapse is a prejudice in arithmetic costume.

8. Measuring it in your own business

1. Baseline. Count lines with a return and the returned quantity from the "Returned" cell — the number usually exists already; nobody has added it up. At the last two stock counts, note the surplus lines with no explanation and their value.
2. Switch it on in the dispatch grid. No separate pilot is needed: returns go through the same grid as deliveries.
3. Re-measure after one month. The month's receipts into the salvage warehouse should be close to returned quantity times cost, and the unexplained surplus at the counts should shrink by about as much.

9. The thing to remember

A number kept only for reconciliation is a number that has not finished its job. The "Returned" cell had been filled in for a long time, correctly and completely. It simply stopped at the warehouse door. The change collects no new data — it lets the data already there travel the rest of its road.

Goods do not vanish when the books miss them. They only lose their name — and come back as a surplus nobody can explain.

The ratios in this piece were measured on a real distributor's system and anonymised; costs and every assumption in the calculator are illustrative. Replace them with your own.