“Classification tells you what type of product something is. Identification tells you exactly which product it is. A good product data strategy needs both — but classification is the foundation that makes everything else work.”
Product classification and product identification serve fundamentally different purposes, and conflating the two is one of the most common mistakes in master data governance. Classification organises products into hierarchical groups — from broad segments down through families, classes, and commodities — enabling spend analysis, search, and procurement workflows. Identification, by contrast, is non-hierarchical: it pinpoints a specific product through a manufacturer's brand name and part number combination, linked to defined property values and units of measurement.
There are several established classification schemes, each designed for different purposes. The three you’ll encounter most often in industrial and procurement contexts are UNSPSC (widely used for spend analysis and procurement), ECLASS (strong in engineering and manufacturing, with rich product properties), and the WCO Harmonized System (required for international trade and customs). Your choice depends on what you need classification to do for your business: enable catalogue search, support spend analytics, comply with trade regulations, or facilitate data exchange with suppliers and customers.
Teams jump straight to Level 4 across all categories, spending months on classification when Level 2 or 3 would answer their questions.
Define the business questions first, then choose the level of depth that answers them.
Schemes update annually. New codes appear, old ones are deprecated. A code without a version reference can mean different things to different recipients.
Always include the scheme version when storing and transmitting codes. Agree on versions with trading partners upfront.
Two "spherical roller bearings" from different manufacturers with different specs look identical in the system.
Classification says what kind of product; identification says which product exactly. You need both layers.
Multi-scheme support. Your solution should handle UNSPSC, ECLASS, ETIM, and the Harmonized System natively — not require a separate project for each.
Version management. Your platform should track versions, manage migrations, and flag deprecated codes automatically.
Scalability. A solution that works for 5,000 SKUs but buckles at 100,000 isn't ready for production.
Standards-based architecture. Look for ISO 8000 compliance — a recognised framework that ensures classified data is interoperable and auditable.
Property-level enrichment. Can the solution attach standardised technical properties (dimensions, materials, tolerances) to each classified product?
Multi-language capability. If you operate internationally, classification needs to work across languages with semantically equivalent terminology.