Why Your Production Count Doesn't Match the ERP

The production count does not match the ERP for four reasons: it is counted by hand or estimated at the end of the shift; it is counted at a different point from the one the ERP uses; scrap and rework are not recorded; and the data reaches the ERP late, aggregated and keyed in by someone else. Each one leaves a small gap. Together they end in shipping documents that do not match what was dispatched, inventories that have to be adjusted at month-end and arguments about which number is the real one.

The good news is that the gap can be measured in a week and is almost always closed without changing the ERP.

Why doesn't what you produced match what the ERP says? #

1. It is counted by hand or estimated

An operator counting boxes, trays or units for eight hours makes mistakes, and not out of carelessness: counting by hand is a task nobody can sustain accurately for a whole shift. When the number is filled in at the end, sometimes it is not even counted: it is estimated from the plan or from the previous shift.

2. It is counted at different points

The line counts at the machine outlet; palletizing counts pallets; shipping counts what leaves on the truck; the ERP records what someone declared. Between one point and the next there are rejects, rework, quality samples and product waiting. If nobody defined which of those points is the official one, every department is right with its own number.

3. Scrap is not recorded

What gets thrown away at startup, what is pulled for quality or what breaks in transit is rarely logged in the same system as the count. That scrap shows up later as an "unexplained" difference.

4. The data arrives late and aggregated

A shift report keyed in the next day delivers a total, not the sequence. If the total does not match, there is no way to know in which hour, on which line or in which lot the difference occurred.

How to measure the gap in one week #

For one week, write down three numbers per line and shift: what the counter or PLC shows, what the manual report says and what ended up in the ERP or on the shipping document. That is enough to see where reconciliation breaks down.

The production KPI spreadsheet has one column for what was counted and another for what was recorded, and highlights in red the shifts that diverge by more than 2 %.

How do you close the gap? #

1. Define the official counting point. Just one, the one that corresponds to what is sold or shipped. The other points help explain the difference, not compete with it.
2. Count on the line, not on paper. With a counting sensor at the machine outlet, or with machine vision on the cameras that already watch the line. The count is stored with time, line and lot.
3. Record scrap in the same system. Rejects, startups and rework with their cause, so the difference stops being unexplained.
4. Send the data to the ERP per event, not per report. Each lot or each pallet with its count, instead of a total keyed in the next day.
5. Review the gap every day. A simple rule: if counted and recorded diverge by more than a threshold, someone reviews it that same day, while it is still possible to find out why.

A real case: PROA #

At Grupo PROA, a poultry producer, counting runs on 19 security cameras that were already installed across six farms. Vision-based counting was measured against manual counting at 99 % accuracy, is integrated with SAP S/4HANA and has been in production for two years. The details are in the PROA case.

What made the difference was not the camera, but counting at the right point and having the number reach the ERP on its own.

Frequently asked questions #

Do you have to change the ERP to make the count match?

Almost never. The ERP is usually fine; what fails is the data it receives. Changing how you count and how the data gets there resolves most of the gap.

Can we count with the security cameras we already have?

Often, yes, if the camera looks at the counting point with reasonable framing and lighting. What determines the result is where the camera is, not its brand. The video is processed on a unit installed in the plant.

What gap between count and record is acceptable?

It depends on the product and the value of each unit, but a sustained gap above 2 % already justifies reviewing the counting process before discussing any other indicator.

If you want to find where reconciliation breaks down in your plant, let's talk.