ETIM 10.0 contains 5,640 product classes and 16,728 features. Neither of those numbers is the one that hurts. The one that hurts is 73,085, the count of class-feature relations in the model. A feature in ETIM carries no fixed unit and no fixed value list. Both are set per class. That single design decision is why ETIM mapping projects fall apart in month two. It is the first thing we walk through on every ETIM engagement.
What ETIM is, and what it is not
ETIM gives you a class per product type and a defined set of features for that class. Values and units come from closed lists. It is a classification model for technical products, maintained by ETIM International and run locally by national organisations in 21 countries. In our market that is ETIM UK and Ireland Ltd, which became the combined body for Great Britain and Northern Ireland in January 2025.
The scope is deliberately technical. Electrotechnical, HVAC and plumbing, building materials, shipbuilding, and the tools that sit alongside them. Cable, switchgear, fixings, boilers, radiators and insulation are all covered. Garden furniture is not.
ETIM is also not several things people assume it is. It is not your website taxonomy. Shoppers do not browse by EC code, and the class structure was never designed for merchandising. It is not a product data model either, because it says nothing about your pricing, your media, your pack hierarchy or your logistics fields. Keeping the classification standard and the merchandising taxonomy and attribution model as two separate layers is the single biggest structural decision in this work. We have covered the general shape of that separation in our piece on B2B product classification.
One more thing worth knowing before you budget. The master English version of the model is free. ETIM publishes it under the Open Data Commons Attribution Licence. National language translations are a different matter and most sit behind membership of that country’s ETIM organisation. The standard is not the cost. The mapping is.
Why UK distributors are being asked for ETIM now
The Electrical Distributors’ Association started pushing electrotechnical ETIM adoption in 2017. A 2019 agreement with the Builders Merchants Federation widened the remit to HVAC, plumbing and building materials. Each body acts as sector lead. EDATA, the UK industry data pool for the electrical sector, is built on ETIM as its base standard.
That is what changed the commercial pressure. Wholesalers pull from the pool. A manufacturer who cannot supply ETIM data gets listed with a name, a price and an image. The competitor next to them gets a full set of filterable technical features. On an electrical distributor site, that gap decides the sale.
The five objects in the ETIM data model
The model is small. Five object types, each with a two-letter prefix and a six-digit number.
- EG, group. A container for classes, for example EG000017. Groups exist to make class management workable. They are not a browsing hierarchy.
- EC, class. The product type, for example EC000016. This is the unit of everything.
- EF, feature. A characteristic, for example EF000138.
- EV, value. A permitted value for a feature, for example EV000147.
- EU, unit. A unit of measure, for example EU000015.
Synonyms hang off classes and carry no identifier of their own. Every class has at least its own name as a synonym, and most carry several. ETIM 10.0 counts 33,684 of them. They are how a search for “consumer unit” finds the class a manufacturer knows as something else entirely.
Features come in four types. Type A is alphanumeric, meaning a closed list of values. Type L is logic, true or false. Type N is a single number. Type R is a range, which needs two numbers. The type is fixed when the feature is created and cannot be changed afterwards. Changing it means a new feature ID, which means a migration.
ETIM 10.0 added feature groups, an optional way to sort a class’s features into sections such as dimensions, electrical and material. That helps the people doing data entry. It does not change the meaning of anything. If you want the wider view on how features map onto a working schema, our guide to product attributes covers it.
Where ETIM mapping breaks, part one: units
The ETIM guidelines put it plainly. A feature does not have a standard unit or value list. The unit or value list for a feature is defined at class level.
Read that again if you are about to write conversion logic. The same feature in two different classes can expect two different units. Any rule table keyed on the feature alone will convert correctly in one class and silently write the wrong magnitude in another. Conversion has to key on the class and feature pair. We have picked this up in audits. A whole range of cable trunking had been loaded in centimetres against a feature expecting millimetres. Nothing flagged it, because every value was a valid number.
Metric and imperial are handled by separate feature codes, not by a unit switch. The guidelines use busbar thickness in millimetres and busbar thickness in inches as two distinct features. If you sell into markets that want both, one supplier column has to feed two features. That is a conversion plus a second write, not a copy.
Countable features carry no unit at all. Number of poles is a number and nothing else, because the unit would be meaningless.
Range features need two numbers and suppliers routinely send one. “Operating temperature 40 degrees” is not a range. Somebody has to decide whether that is the upper bound, the nominal figure or a typing error. That somebody should be a product person, not a developer.
Then there is rounding. The model does not set your rounding policy. Agree significant figures per class, in writing, before a single value is enriched. Otherwise you get 2.5, 2.50 and 2.500 in the same feature and your faceted search splits into three.
The honest version of this problem is not the maths. It is that the 700-column supplier XML contains “2.5mm2”, “2,5 mm²” and “2.5” in the same column. Unit conversion cannot tell you which of the three the supplier meant.
Where ETIM mapping breaks, part two: value lists
Type A features are closed lists, and those lists are defined per class. Suppliers send free text. Every ETIM programme therefore needs a persistent mapping table from raw supplier strings to EV codes, and that table is the real asset. The classified output can be regenerated. Two years of resolved supplier strings cannot.
The guidelines add a rule that causes more argument than any other. Value lists should be unambiguous and of the same level. Plastic and Polyethylene (PE) should not appear in the same list, because PE is a plastic. That is correct as a modelling principle and awkward in practice. When a supplier writes “PE” against a class whose list contains only Plastic, you have a judgement call, not a lookup.
Two rules we hold to on this. First, resolve at the class level and record the decision, so the next person inherits it rather than repeating it. Second, never force a near match. A wrong EV code is worse than a blank one. A blank shows up in your completeness report and gets fixed. A wrong value looks complete, passes every check, and comes back as a return.
Most of that work belongs upstream in the supplier feed, not downstream in the PIM. Our supplier data onboarding approach exists mostly because of this one problem.
Where ETIM mapping breaks, part three: class migration
Class versions are no longer tied to release versions. A new class starts at version 1 and the version only moves when that class actually changes. So “we are on ETIM 10” tells you nothing useful about which version of EC000016 your data was mapped against. Record the class version per class, not the release version per catalogue.
Retired classes and features are deprecated rather than deleted. They stay in the tables because live data still points at them, and they disappear from selection rather than from existence. Dynamic releases between official ones carry the full official release plus everything tagged for deletion next time round.
Splits and merges are where the effort lands. The guidelines describe specialisation, where a Hammer class is retired in favour of claw hammer, smith hammer and chisel hammer. They also describe generalisation, where several classes collapse into one. Predecessor and successor relationships are recorded, which helps. What that table cannot tell you is which of the three hammer classes any given SKU belongs in. That decision is per SKU, and at 40,000 SKUs it is a project.
Plan for the next one now. ETIM 11.0 has a final release date of 1 December 2026. The change request deadline passed on 30 June 2026 and a beta is due in November for testing. If you are mapping this year, map against 10.0. Treat the 11.0 delta as its own scoped piece of work, not a patch dropped in over Christmas.
Getting the data out: IXF, BMEcat and ETIM xChange
Three formats get confused with each other, so it is worth separating them.
IXF is how ETIM distributes the model itself. ETIM 10.0 shipped with IXF 3.1. You consume it. You do not publish in it.
BMEcat is the exchange format that carried ETIM product data for years. It still carries plenty of it, particularly in Germany and the Netherlands. If your customers ask for a BMEcat file, they mean product data with ETIM classification embedded.
ETIM xChange is ETIM’s own exchange standard, it is JSON, and it is now the recommended route. Version 2.0 was released on 27 November 2025. It added life cycle assessment and environmental product declaration fields. It also added packaging material data for the EU packaging rules, plus serial number handling.
That last release tells you where this is heading. Sustainability and compliance data now arrives down the same pipe as technical data. That is the same pattern we see in digital product passport work. If your ETIM data lives in spreadsheets, you can produce a file once. You cannot maintain one through two more format versions.
How we sequence an ETIM mapping
Six steps, in this order, every time.
- Rank classes by revenue, not by SKU count. The long tail of 4,000 slow movers is not where the return is. In most distributor catalogues we see, a few hundred classes carry the bulk of the trade.
- Assign classes before touching features. Class assignment is a merchandising judgement. Feature population is a data job. Mixing them means the data team makes classification decisions nobody signed off.
- Fix the unit and rounding policy per class, in writing, before enrichment starts. This is a half day of work that saves a re-run.
- Build the supplier string to EV mapping table on day one. Even if it starts with six rows.
- Populate features in waves. Mandatory technical features first, the ones buyers actually filter on second, the rest when there is capacity.
- Measure completeness per class, not per catalogue. A catalogue at 78 percent complete can hide a class at 12 percent, and that class is usually the one your biggest customer searches.
We work with industrial and electrical distributors including RS Group and APS Industrial. The shape of this problem is the same at every size. The same sequencing applies whether the catalogue holds 20,000 SKUs or 700,000. It is also why the taxonomy and attribution layer has to be settled before ETIM population starts, not after. For the wider picture on getting technical specifications right, see our piece on industrial product data and technical specs.
Key takeaways
- ETIM 10.0 holds 5,640 classes and 16,728 features, but 73,085 class-feature relations. Units and value lists belong to the relation, not to the feature.
- Metric and imperial measurements are separate feature codes. Convert per class and feature pair, never per feature.
- Type A value lists are closed and class-specific. The supplier string to EV mapping table is the asset worth versioning and keeping.
- Class versions move independently of release versions. Record the class version against each product.
- ETIM 11.0 lands on 1 December 2026. Map against 10.0 now and scope the delta as separate work.
- The master English model is free under an Open Data Commons Attribution Licence. Your cost is the mapping effort and the ongoing maintenance.
Where to start
You are being asked for ETIM data and you are not sure what the answer is. A mapping exercise, a PIM change or a supplier onboarding problem? Thirty minutes with someone who has done all three is usually enough to tell. We run a short discovery call, look at a sample of your supplier data, and tell you which one you face.
Book a thirty minute discovery call, or read more about how we approach PIM and PXM services for distributors.