Should we design our product data model before selecting a PIM?

Design the product data model before selecting a PIM, because the data model is what every vendor will be tested against. The core finding behind the PIMvendors method is that most PIM selections fail before the first demo, when buyers compare software before they have defined their data model.

A selection-ready data model covers 6 elements:

• Product families and classification, including standards such as ETIM or GS1.
• Attributes per family, with data type, unit, validation rule and language dependency.
• Variant logic, for example size, colour or configuration.
• Relations: accessories, spare parts, replacements and bundles.
• Assets linked to products, with their metadata.
• Ownership per attribute group, plus the source system for each attribute.

The model can stay partial at this stage. PIMvendors estimate: 2 or 3 representative product families, modelled in full and including the most complex one, are enough to write requirements and demo scenarios for a mid-sized selection.

This follows the PIMvendors sequence: Data, then Process, then People, then Tooling. A data model built first gives vendors a fair test, exposes where a platform's modelling stops, and shortens implementation because the core design work is already done.

Use the sample families as demo data, and send them to every shortlisted vendor at least one week before the demo.