Part classification with eCl@ss, ETIM and UNSPSC: which standard, and how to roll it out
Free-text descriptions turn search, comparison and parts shops into guesswork. How eCl@ss, ETIM and UNSPSC differ, which fits a machine builder, and a rollout that doesn’t stall after the first 1,000 parts.
Not ready to talk yet? Take the 3-minute Data Foundation Check
TL;DR
- A class code alone isn’t the goal. The value sits in the attributes: diameter, voltage, material, thread.
- eCl@ss is strong in German-speaking industry, ETIM in electrical parts, UNSPSC in spend reporting.
- Many machine builders use two: eCl@ss or ETIM for product data, UNSPSC for spend. A mapping table connects them.
- Start where it pays: standard parts with many duplicates and spare parts customers search for.
Why classify parts at all?
A description like “BRG 6204 2RS SKF” works for the person who typed it. For everyone else, and for every system, it’s noise. Classification turns that text into a class plus attributes that can be searched, compared and checked.
| Free text | Classified |
|---|---|
| BRG 6204 2RS SKF | Deep groove ball bearing · inner Ø 20 mm · outer Ø 47 mm · width 14 mm · seal 2RS · maker SKF |
| Bearing 6204 ZZ | Deep groove ball bearing · inner Ø 20 mm · outer Ø 47 mm · width 14 mm · shield ZZ |
| Kugellager 6204 | Deep groove ball bearing · inner Ø 20 mm · outer Ø 47 mm · width 14 mm · seal unknown |
Once the attributes exist, duplicates show up on their own, buyers can bundle demand, and a parts portal can offer filters instead of a search box that only finds exact matches.
eCl@ss, ETIM or UNSPSC: what’s the difference?
| eCl@ss | ETIM | UNSPSC | |
|---|---|---|---|
| Maintained by | ECLASS e.V., Germany | ETIM International | Global code set, procurement focus |
| Strength | Deep property sets for industrial parts | Detailed features for electrical and HVAC products | Simple hierarchy for spend analysis |
| Structure | 4 levels + standardised properties | Classes + features | 4 levels, few attributes |
| Typical use | Product data, PIM, catalogues, customer data requests | Electrical components, wholesale data exchange | Spend cube, procurement reporting |
| Fits machine builders for | Mechanical and electrical parts | Control cabinet and electrical parts | Spend analysis across categories |
A shortcut that backfires
Assigning only the class code and skipping the attributes. It looks finished on a dashboard, but nobody can search or deduplicate on a class code alone.
How do you roll it out? 5 steps
- Pick the standard and version, and freeze it. Switching versions halfway creates mapping work later.
- Define an attribute subset per class. A standard class can carry dozens of properties. Pick the 5 to 10 that matter for search, duplicates and service.
- Extract from what you already have. Descriptions, drawings, supplier catalogues and manufacturer part numbers. Rules and text parsing cover a large share. People review the rest.
- Validate with units and value lists. “20”, “20mm” and “Ø20” are the same value. Normalise before you load.
- Build it into the new-part process. Classifying at creation takes seconds. Fixing it later takes projects.
What attribute normalisation looks like
RAW NORMALISED RULE "20mm" 20 mm split value and unit "Ø 20" 20 mm strip symbol, default unit "0,02 m" 20 mm convert unit "M8x1.25" M8 · pitch 1.25 thread pattern
Where should you start?
| Part group | Why first | Effort |
|---|---|---|
| Fasteners and standard parts | Most duplicates, easy attributes | Low |
| Bearings, seals, filters | High service relevance, clear dimensions | Low to medium |
| Electrical components | Rich manufacturer data available | Medium |
| Purchased assemblies | Needed for spend and risk analysis | Medium |
| Machined drawing parts | Class is easy, attributes sit in drawings | High |
Checklist
- Standard and version chosen and frozen
- Attribute subset defined for priority classes
- Manufacturer part numbers available for purchased parts
- Unit and value rules written down
- Classification is a required step for new parts
FAQ
What is eCl@ss?
eCl@ss (also written ECLASS) is a classification standard for products and services maintained by ECLASS e.V. in Germany. It provides a four-level class hierarchy with standardised properties and is widely used in German-speaking industry.
What is the difference between eCl@ss and ETIM?
Both classify products with attributes. ETIM comes from electrical and technical wholesale and is strongest for electrical, HVAC and installation products. eCl@ss covers a broader industrial range.
Is UNSPSC enough for product data?
Usually not. UNSPSC is good for grouping spend, but it carries few attributes. For search, duplicate detection and catalogues you need an attribute-rich standard such as eCl@ss or ETIM.
Can we use more than one standard?
Yes, and many companies do. Keep one leading standard for product data and maintain a mapping to the others for reporting or customer requirements.
How many attributes should we fill per part?
Only those that drive search, duplicate detection and service. For most classes that means 5 to 10 attributes, not the full property list.
How much of your part master can be classified automatically?
Send us a part master extract under NDA. We’ll show you the share we can classify by rules and what needs a human.





