Skip to content
allanninal.dev/build
Collaborate

Part 5 of 7 · Packing slip checker series ~5 min read

How discrepancies become a pattern

Individual discrepancies get resolved and forgotten, which is fine for each one and terrible in aggregate, because the recurring ones are where all the money is.

Key takeaways

  • Group by supplier, by product, and by unit of measure. The third one finds the most.
  • A supplier consistently short by a small percentage is usually a systematic cause.
  • Track the acceptance rate too: lines never counted cannot show up as discrepancies.
  • Report the value, not the count. Eleven small shortages may matter less than one large.
  • The output is a supplier conversation, not a scorecard.

Three groupings

Three ways goods receipt discrepancies are grouped for analysisThree boxes stacked on the left. By supplier: who is short and how often, labelled obvious. By product: one item across many suppliers, labelled often packaging. By unit of measure: boxes versus units, labelled finds the most. All three converge on A quarterly view with counts and value, and that leads down to One conversation each, with evidence attached. A note says the third lane is unglamorous and repeatedly turns out to be the largest cause.By supplierwho is short, how oftenobviousBy productone item, many suppliersoften packagingBy unit of measureboxes vs unitsfinds the mostA quarterly viewcounts and valueOne conversation eachwith evidence attachedThe third lane is unglamorous and repeatedly turns out to be the largest cause.
Fig 1. The three ways discrepancies are grouped. The unit-of-measure grouping is the one nobody thinks to build and the one that finds systematic errors.
  • Analytics
  • Front-end & mobile
  • People

Unit of measure is where the money is

The classic case: the order says sixty units, the supplier’s system holds the product in boxes of twelve, and somebody enters five. Five boxes is sixty units and everything is correct. Then the packaging changes to boxes of ten and nobody updates the mapping, and every order is short by ten units from that day onwards until somebody counts.

That error is invisible to a supplier-level analysis, because the supplier is only short on one product. It is invisible to a product-level analysis if the product comes from several suppliers. It shows up immediately when discrepancies are grouped by the unit mapping in use.

The acceptance rate belongs in the report

The proportion of lines physically counted for three suppliersA stacked bar chart with three bars in per cent. Two series: lines physically counted in green, and lines accepted on the slip in grey. Supplier A shows sixty-two per cent counted and thirty-eight accepted. Supplier B shows twenty-four counted and seventy-six accepted. Supplier C shows eleven counted and eighty-nine accepted. A note says Supplier C has almost no discrepancies, and almost nothing from them has been counted.050100150200~100Supplier A~100Supplier B~100Supplier CLines physically counted, %Lines accepted on the slip, %Supplier C has almost no discrepancies. Almost nothing from them has been counted.
Fig 2. Why the discrepancy report needs the acceptance rate beside it. A supplier with a clean record and an eighty-nine per cent acceptance rate has a clean record for a reason that is not about their accuracy.

This is the correction that keeps the whole report honest. Discrepancy counts are counts of things that were looked for, and a supplier who is rarely checked will always look reliable. The rotating spot-check from Part 2 exists partly to keep this from getting extreme, and the acceptance rate is how you know whether it is working.

Value, not count

Eleven shortages of two units each on a cheap consumable is worth less attention than one shortage of four units on something expensive, and a report ordered by count puts them the wrong way round.

Ordering by value also makes the report shorter, which is the property that determines whether anybody reads it. Three lines with money attached beats forty lines with counts.

What the report says

The quarterly page

  • Total value of discrepancies: £4,180 across 34 deliveries, of which £3,020 recovered.
  • Largest single cause: a unit-of-measure mapping on one product, 9 occurrences, £1,340.
  • By supplier: three named, with counts, values and their acceptance rates.
  • Acceptance rate overall: 71% of lines were not physically counted.
  • Over-shipments: 6, value £510, all reported and 4 collected.
  • Unresolved after 60 days: 2, both with the same supplier.

The second line is the one that pays for the system. A mapping error found in a quarterly report and fixed in five minutes was quietly costing more per year than the entire rest of the discrepancy list.

The last line matters too, and it is the one most reports omit. Discrepancies that were raised and never resolved are the ones that get written off, and they are invisible unless something is counting how long they have been open.

A conversation, not a scorecard

As with lead times, the output is evidence for a conversation rather than a grade. “Nine of the last thirty deliveries were short on this one product, here are the photographs, we think it is the box size mapping” is a productive opening. A supplier accuracy percentage is not, because there is nothing in it to discuss.

Next: what all of this costs to run.

All posts