Invoicing at the original PO price: the delivery note says 21,000, the chain's PO says 30,000

A retail chain pays the price on its own PO, not the price on the supplier's delivery note. When a machine reads the PO price, invoices at it when the gap is plausible and corrects the price list for next time, what is that worth a month — and which lines still need a person?

A delivery note to a retail chain lists tray-packed guava at 21,000 VND/kg. The chain's purchase order (PO) for that same item says 30,000 VND/kg. The accountant invoices by hand at the PO price — 30,000 — and that is correct: the chain pays by its own PO, not by the supplier's paperwork. But the figure is right only because a person opened the PO and read it. Once invoicing moves to a machine that invoices from what the supermarket received, the machine will take the price on the delivery note, and every kilogram of that guava will be 9,000 VND short on the invoice.

This article asks: if the machine reads the original PO price itself, invoices at that price when the gap is reasonable, and corrects the price list for next time, what is that worth each month — and which part still has to be left to a person?

1. Where the price gaps come from

The chain's prices do not stand still. Measured over ten days of POs from one chain, the same item changes price by batch of days, not by store: one boxed rose apple was 57,000 VND/kg from the 20th to the 24th, 56,000 from the 25th to the 27th, then 52,000 from the 28th. Meanwhile the supplier's price list changes only when the sales team remembers to change it. Worse, over 30 days, 1,057 of 1,139 sales orders for one group of the chain's stores carried the system's default price list instead of the customer-group price list — because the order screen pre-fills the default list. Delivery notes printed from those orders carry the prices of a different list.

2. What changes

1. The machine reads the PO price for each item. The chain's goods receipt has no price, so the machine follows the PO number on the goods receipt → the list of purchase orders → the order detail, and writes the PO unit price onto each goods receipt line.
2. Same unit, and the PO price between ½ and 2 times the delivery note price ⇒ the invoice is made at the PO price, labelled «At PO price», with the old delivery note price shown alongside. The note still goes straight through; nobody has to click anything.
3. Different unit, or too large a gap ⇒ «Price gap», waiting for a person. Measured examples: PO 85,000 VND/kg against a delivery note at 550,000 VND/carton — both figures are correct, they are simply two different things. Or 33,000 VND/kg on the delivery note against 78,000 on the PO, under the same word «kg»: a 500 g box priced per box on one side and per kg on the other.
4. The customer-group price list corrects itself to the latest PO price, under the same unit and ½–2× conditions, and every corrected line records its originating PO and PO date. Because prices change by batch of days, the machine takes the price of the most recent PO date (if that day has several prices, the one that appears most often), not «the most common price over seven days» — that method returned exactly the old price.
5. New orders take the customer's price list, not the system's default list.

The ½–2× range is a number set by a person, and it alone decides where the money goes. The figure below lets you drag it.

3. What the first four days measured

On the invoices the machine made for one chain from delivery notes dated 26 to 29/10: 40 of 88 invoices had at least one line at the PO price — 61 lines in all. On those 61 lines, the PO price was higher than the delivery note by a total of 1,250,300 VND (money that would have been under-invoiced) and lower by 1,557,500 VND (money that would have been over-invoiced, then short-paid by the chain and left to reconcile). Both directions together come to 2.8 million VND on 90.3 million VND of invoices over four days — about 3.1%.

On a wider sample — 1,000 goods receipt lines that had a PO price and were matched to a delivery note line, from 17 to 30/10, two chains — 934 lines matched on price, 66 did not. Of the 66 with a gap, 45 were the same unit and within ½–2×, and the machine invoiced them itself; 21 waited for a person: 11 with the same unit word but a gap of more than 2×, 10 with a different unit. The two chains differ widely: one almost always matches, the other has a gap on close to three quarters of its lines. Measure your own chain; do not use the average.

4. Source one: revenue no longer under-invoiced

How to measure. Take one month of hand-made invoices, compare each line with its PO, and count the lines invoiced below the PO that were never made up. That is your own company's miss rate — the 30% above is an illustration.

Confidence: fair. The gap money is measured; the miss rate has to be measured, and with a careful accountant it can be close to 0.

5. Source two: lines no longer short-paid and then reconciled

The other direction does not lose revenue — the chain pays by the PO whatever the invoice says — but every over-invoiced line is one receivable that does not match, one reconciliation, one adjustment invoice.

How to measure. Count the short payments from the chain over three months whose cause was the unit price, and ask the person who reconciles how long one takes.

Confidence: moderate. The number of short payments can be counted; the minutes per payment vary widely by chain.

Each line the machine corrects sits in only one of the two sources above — either the PO is higher, or it is lower. Nothing is counted twice.

6. Source three: no opening the PO to compare each line

How to measure. Time one session of invoicing the chain the way it is done now, and divide by the number of lines. If your company has never compared against the PO price — that is, it invoices at the delivery note price — this source is 0 and source one is much larger.

Confidence: high. This is the largest source and also the easiest to measure.

These hours are money or capacity for the accountant to do other work — count them once.

7. What is left for people

Narrow the range and this part grows; widen it and the machine starts applying prices to lines that are in fact a different pack size. In the sample, no line with the same unit fell between 2 and 2.3 times — in this data, that empty gap is where to put the edge.

8. Adding it up

About 3 million VND a month, 37 million a year at 3,000 lines. Two thirds of that is time spent reading POs, not gap money. That says something real: in a company with a careful accountant, the value of this lies in no longer needing that care on every line; in a company where nobody compares against the PO price, the value lies in revenue.

9. What assumptions this ROI rests on

The PO price is the price the chain will pay. All four sources rest on that sentence. If a chain pays a different price — by a contract price list, by a promotion settled later — then invoicing at the PO price only moves the gap somewhere else.

The ½–2× range does not catch every pack-size mismatch. A 500 g box priced per box on one side and per kg on the other differs by exactly 2× if the per-kg price is stable — landing right on the edge of the range. That is why every line the machine applies itself carries a label, and the reviewer can see the old price next to the new one.

A price list stays put only when no other path writes over it. On the first day, the machine corrected 89 price list lines across 9 customer-group price lists (79 lines changed price, 10 lines had no price before; median gap 15%). But 42 times, a line the machine had just corrected was changed back before the next run. A sales screen that wrote prices straight into the list was fixed the same day; the number to watch from here on is the number of price list lines the machine has to correct a second time — it must come down to near 0.

Correcting the price list adds no money line to the calculation. It moves the price gap from «corrected at invoicing» to «no longer happens»: the share of lines with a price gap next week must be lower than this week. If it is not lower, the price list is not helping.

10. Measure it in your own company

1. Take a baseline. One month: invoice lines per chain, lines where the PO price differs from the delivery note, hours spent invoicing chains, short payments from chains caused by the unit price.
2. Run with the ½–2× range for four weeks, and each week record: lines the machine invoiced at the PO price, «Price gap» lines and how the person decided, price list lines corrected a second time, the share of lines with a price gap.
3. Compare each chain with itself. The two chains in the sample had gaps on 1% and close to 75% of lines; pooled together, the number says nothing.

Further reading: if the quantity on the invoice must also follow what the supermarket received, the article signed delivery note checked the same day covers that part.

11. What to remember

There is one price the customer will pay, and it is on the customer's paper. Every time your own system invoices at your own price without asking the customer's paper, the price gap becomes the month-end reconciler's problem. Read the PO price at the moment of invoicing, let the machine correct what falls neatly within a range a person has set, and send the rest to a person — with both figures.

The right unit price is not the unit price in your own list. It is the unit price the person paying has written down.

The number of lines invoiced at the PO price, the gap money in both directions, the 1,000-line sample and the number of price list lines corrected were measured on the live system of a fresh-produce supplier to retail chains on 30/10/2026, anonymised. Lines per month, miss rate, minutes and wages are illustrative figures. Replace them with your own.