Composite pattern: a PIM selection driven by packaging compliance

Situation: packaging data held by engineers and suppliers

This is a composite pattern built from several PIMvendors selections, with no client or vendor named. It describes a European consumer goods producer with a broad range sold through retailers in several EU countries. The EU Packaging and Packaging Waste Regulation applies generally from 12 August 2026.

Packaging data sat with packaging engineers, procurement and suppliers. Material composition, component weights and recyclability information lived in spreadsheets and supplier declarations. Product data for retailers sat in a separate system with no link to packaging.

The cases overview lists the other composite patterns.

Complication: the compliance owner did not own the data

The compliance team owned the obligation, while the data belonged to three other departments. Packaging engineers held the specifications. Procurement held supplier declarations. Marketing held the product records that retailers received.

Packaging was shared across products. One bottle design served many products, and one product used a primary, secondary and transport layer. Spreadsheets could not hold that many-to-many relationship reliably.

Evidence was the third gap. Supplier declarations arrived as PDFs with no link to the values typed into spreadsheets. An auditor could not trace a reported value back to its source.

What the selection did differently

The selection modelled packaging as its own entity before asking vendors anything. The order followed Data, Process, People and Tooling.

1. The team modelled packaging components, materials and weights as a separate entity linked to products.
2. Each regulatory attribute was given an owner, an evidence requirement and an approval step.
3. The supplier declaration process was mapped, including how evidence documents attach to values.
4. Requirements were drawn from the regulatory product data requirements, tagged across all four domains.
5. AI was limited by policy to extraction drafts, with human approval for every regulatory value.

Which scenarios decided the selection?

The decisive scenarios tested structure, evidence and traceability on real packaging data.

A fifth scenario tested AI extraction from a supplier PDF. Reviewers checked each extracted value against the document during the demo.

Outcome patterns

Entity modelling separated the shortlist more clearly than any other capability. Vendors with flexible entity models handled shared packaging natively. Vendors limited to product attributes needed duplication or custom development, which made the data harder to keep consistent.

AI extraction was useful as a first draft and inconsistent on complex declarations. That result matched the house position that AI remains unreliable for regulatory data. The approval step stayed in the design.

The ownership work outlasted the selection. Naming owners for each regulatory attribute gave the compliance team a clear route to the data it needed.

Where this pattern goes wrong

The pattern fails when packaging data is handled as a one-off compliance project. A spreadsheet built for the first report goes stale as soon as a supplier changes a material. Without a process and an owner, the next report starts from scratch.

Leaving the compliance team off the demo panel is a second failure point. Packaging engineers judge the model, compliance owners judge the evidence trail, and both views matter. A third risk is accepting AI-extracted values without review. Extraction speeds up data entry, and each regulatory value still needs a person to approve it against the source document.

Lessons for producers in a similar position

Model packaging and assign owners before comparing software. Four lessons recur across the engagements behind this pattern.

• Model packaging as a separate entity linked to the products that use it.
• Attach evidence to values so an auditor can trace each one.
• Keep human approval for every regulatory value, including those drafted by AI.
• Track regulatory dates on the Compliance Countdown and plan requirements around them.

Groups replacing an existing PIM while preparing for PPWR can compare this pattern with the multinational replatforming pattern. The manufacturer pattern shows the same data model discipline for technical products.