What is a SKU?

Last updated: August 2026

A SKU (stock keeping unit) is an internally assigned identifier that retailers and manufacturers use to distinguish every sellable unit in their assortment — down to the individual variant in size, colour or style. Unlike an EAN or GTIN, a SKU is not issued by a central organisation; each company defines its own SKU logic.

What does SKU stand for?

SKU stands for “stock keeping unit”: the smallest distinctly identifiable unit a company stocks, sells or tracks — and, in everyday use, the code that identifies exactly that unit. When someone asks for the “product SKU”, they mean this identifier for one concrete product in one concrete configuration.

The level matters: a t-shirt model is one product, but “t-shirt, blue, size M” is one SKU and “t-shirt, blue, size L” is another. Anything you need to count, stock or analyse separately gets its own SKU — which is why the SKU count of an assortment is almost always far higher than its product count.

Is a SKU the same as an EAN or GTIN?

No. A SKU is an internal identifier that every company assigns according to its own logic — an EAN or GTIN is a global standard governed by GS1. A GTIN (Global Trade Item Number) identifies a trade item uniquely worldwide and is derived from a company prefix licensed from GS1; in 2009, the term GTIN officially replaced the older name EAN (European Article Number) — both refer to the same number.

The practical consequence: the same product carries a different SKU at every retailer, but the same GTIN everywhere. Marketplaces and trading partners usually require the GTIN, while your inventory and internal processes run on the SKU — so both identifiers must be reliably linked in your systems.

What is the difference between SKU, UPC, GTIN and MPN?

Each identifier answers a different question — mixing them up produces duplicates, broken feeds and rejected marketplace listings:

In product data management, this is a core discipline: every article and every variant carries all relevant identifiers as maintained, validated attributes — the SKU as the internal key, GTIN and MPN as external references. A missing or duplicated mapping shows up immediately as a data quality problem in your channels.

How is a SKU structured? (Example)

A good SKU is short, unique and follows a fixed scheme that people can read and systems can sort. A common approach uses meaningful segments — for example category, model, colour, size.

Example for an apparel range: TS-CLASSIC-BLU-M = t-shirt, model Classic, blue, size M; its neighbour in size L is TS-CLASSIC-BLU-L. Anyone who knows the scheme reads the code at a glance, and related variants sort next to each other in every export.

Proven rules: start with the most distinctive attribute; avoid special characters, spaces and leading zeros that systems interpret differently; prefer letter–number combinations over long digit strings; never encode volatile information such as prices or suppliers into the SKU — that belongs in product attributes, not in the key.

How do you create an SKU system?

An SKU system takes four steps: choose the attributes, define the scheme, document it, enforce it.

1. Choose attributes: Which properties genuinely distinguish your stock units (category, model, colour, size, pack size)? Only those belong in the scheme.
2. Define the scheme: Fix segment order, lengths and allowed characters — consistently across the whole assortment, with room for future attributes.
3. Document it: Write the scheme down and brief every role that creates articles. SKU systems rarely fail on logic; they fail on unmanaged assignment.
4. Enforce it technically: The leading system — usually ERP or PIM — enforces uniqueness, validates the format and blocks duplicates. Manual assignment in spreadsheets is the most common source of SKU chaos.

Once assigned, SKUs should stay stable: systems, feeds, warehouse locations and analytics all depend on them. Changing the scheme later is possible, but it is a migration project, not a renaming exercise.

How do you assign SKUs to product variants?

Every sellable variant gets its own SKU, derived from the parent product’s logic — typically a base code plus variant segments. That keeps it visible which SKUs belong to one product while every configuration remains individually countable and orderable.

This is where assortments scale or collapse: with three colours and five sizes, a single shirt model already has 15 SKUs. How to model variant structures — master variant, variant tree, inherited versus variant-specific attributes — is covered in depth in What is variant management? For SKU assignment, the rule is: the code structure should follow the variant structure in your system, not the other way round.

Why are SKUs important in e-commerce and warehousing?

The SKU is the key that connects inventory, processes and analytics. In the warehouse it identifies the unit at its location: picking, stocktaking and replenishment all work at SKU level.

In e-commerce, practically everything hangs on it: the shop tracks stock and variants per SKU, marketplace feeds reference it, returns are booked against it, and metrics such as sell-through, return rate and stock cover are analysed per SKU. Managed well, SKUs let you distinguish best-sellers from shelf-warmers at variant level — managed badly, you only ever see product-level averages.

How do you manage SKUs in a PIM system?

In a PIM system, the SKU is the unique key that all product information of a unit attaches to: attributes, copy, media, classifications and the other identifiers such as GTIN and MPN. The PIM enforces the rules — uniqueness, format, mandatory fields — and connects the SKU level to the product and variant structure.

The payoff comes at syndication time: each channel receives exactly the identifiers and data it requires per SKU — the shop gets the SKU with stock logic, the marketplace the GTIN, the data sheet the MPN. The foundation is disciplined maintenance of these mappings; the basics are covered in What is product data management?

Where does the SKU sit in the PIM/PXM/DAM landscape?

The SKU is the lowest identification level of product data: the PIM system maintains the complete, verified record per SKU and links it to media from the DAM; PXM processes deliver this data to every channel. ERP and warehouse management use the same SKU for stock and logistics — making the identifier the bridge between merchandise management and product communication. The fundamentals are covered in What is a PIM system?

How does a PIM system like OMN support SKU management?

OMN (Online Media Net) by apollon manages products, variants and their identifiers in one central structure: every SKU carries its attributes, copy, media and the associated external numbers such as GTIN and MPN as maintained fields. Quality and completeness checks ensure that only correct, complete records are delivered to your channels, the variant logic keeps related SKUs structured together, and workflows govern who creates and approves articles. At syndication, OMN delivers the right identifiers per channel — from shop feeds to marketplace listings. apollon contributes more than 25 years of experience in product data communication. See how OMN compares in the PIM software comparison, or request a demo.

FAQ — Frequently asked questions about SKUs

What is a SKU on Amazon or Shopify?

The same principle: your own internally assigned code per listing. Amazon calls it the seller SKU and additionally assigns its own platform-internal identifier (ASIN) per product. In shop systems such as Shopify, the SKU is a field per variant that inventory and reporting run on.

Can you change a SKU later?

Technically yes, practically only as a migration: warehouse locations, feeds, integrations and historical reports all reference the old code. Start with a stable scheme; if you must change it, map old to new SKUs in every connected system.

How long should a SKU be?

As short as possible, as long as necessary — around 8 to 16 characters works well in practice. Uniqueness and a consistent scheme matter more than length; very long codes invite typing errors, very short ones leave no room for meaningful structure.

Does every SKU need its own GTIN?

Every unit that is sold as an independent trade item needs its own GTIN — at variant level that usually means one GTIN per SKU. Internal SKUs without retail relevance (such as warehouse-only bundles) can do without one. The SKU-to-GTIN mapping is maintained in the leading system.