What doesn't a converter's ERP tell you?

What was really run, and what it really cost. A packaging converter's ERP records the order, the specified grade, the standard cost, the shipment and the invoice. Where it costs at standard, it may not record the grade actually run when a substitute was used, the trim and set-up waste on that order, or the handling, freight and service that one customer needed and another didn't. Each of those changes margin, and each sits somewhere else: in the corrugator's records, on the floor, in logistics or in someone's head. So for a converter costing at standard, the figure the ERP reports least reliably is margin by order and by customer.

What the usual answer says

The usual answer is that a converter's ERP lacks machine data: real-time speeds, downtime, quality measures and shop-floor detail. It recommends manufacturing execution systems and sensors to fill the gap. Those systems help run the plant. They do not, on their own, put the cost of what happened back onto the order.

Substitution and trim

In corrugated, the grade on the order is not always the grade on the machine. Constantine Goulimis and Gastón Simone work at Greycon, which sells planning and trim-optimisation software to paper and packaging companies. They note in a 2020 study of reel stock policies for an integrated paper packaging company that "material substitution is common practice in the industry". Their example is running 200 g/m² paper for an order that specified 180 g/m², "giving the end customer a slightly higher specification". They add that "The additional cost may be offset by other factors such as reduced trim waste or inventory / logistics costs".

An illustrative five-order example in the paper shows how sensitive waste is to such choices. Changing one stocked roll width by 75 millimetres moved corrugator waste on the same orders from 3.071% to 2.584%. The authors also observe why plants over-stock widths: "Because side trim is such an easily measured factor, it generally dominates in the stock size selection process." The costs that are harder to measure, inventory and lost corrugator time, get less weight.

An ERP that costs at standard doesn't see this. It costs the order at the specified grade and a standard waste allowance. The upgrade and the actual trim show up, if at all, as a variance spread across the month.

Cost to serve

The second gap is service. Plante Moran, in a July 2025 analysis of cost to serve, lists costs that vary by customer, including ordering by EDI, fax or email, dedicated equipment and overtime. On where that data lives, it is direct: "Although you may be collecting the data, it probably doesn't reside entirely within your ERP or any other single system."

Its example is a consumer-packaged goods company, not a converter. It offered unique displays to its customers. It found that "'Easy' customers were subsidizing those with more service requirements". Repricing on cost to serve led to a 12% increase in EBITDA. That is one case reported by an adviser, in another industry, and it shows the direction more than the size.

For a converter, the same pattern is familiar: short runs with their own dies and plates, rush orders that break the schedule, customers who want stock held. The ERP invoices them at the quoted price and costs them at standard.

What this leaves out

The obvious gap, estimate against actual for makeready and run speed, is left aside here on purpose. Many converters already review it job by job. Substitution and cost to serve get less attention because the data for them sits outside the production system as well as outside the ERP.

How to find the gap

Pick twenty recent orders across customers and run lengths. For each, rebuild margin from what happened: the grade run, the actual trim and set-up waste, the freight, any special handling or storage. Compare it with the margin the ERP reported, and sort the orders by the size of the gap.

AI helps with the rebuild, which is slow by hand. The key is the order or job number. A model can read it from corrugator reports, freight bills and emailed changes and attach each record to its order. It fails where a record carries no job number, such as a freight bill for a combined load. Those need a rule for splitting cost, set by a person.

When it doesn't apply

Converters that already capture actual material and waste per order, as some industry-specific systems do, and cost from them rather than from standards, will find a smaller gap. Plants running long, repeat orders on a few grades have less substitution and less set-up to allocate. And the 2020 study describes one integrated company's stock policy, so its percentages illustrate sensitivity, not typical waste.

Quarri for pulp, paper and packaging is built around how a fibre operation runs, from furnish to converted order.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.