What is a SKU? Definition, structure and common mistakes
The SKU is the identifier everything else hangs off — stock, orders, channels, reporting. Which is exactly why a sloppy scheme causes problems for years.
- ✓A SKU is your internal identifier for a specific, sellable product variant.
- ✓It is not an EAN/GTIN: the SKU is yours, the GTIN is global and usually comes from the manufacturer.
- ✓Every separately sellable variant needs its own SKU — size and colour included.
- ✓Keep the scheme short, consistent and stable; renaming later is disruptive.
SKU stands for Stock Keeping Unit. It is the identifier you assign to a specific, sellable product variant so your own systems can tell it apart from everything else you sell. Not the product line, not the model — the exact thing a customer can put in a basket and you can count in a warehouse.
SKU vs. EAN/GTIN — the distinction that matters
These two get mixed up constantly, and the difference is simple: the SKU is yours, the GTIN is the world's.
| SKU | EAN / GTIN | |
|---|---|---|
| Assigned by | You | GS1 system, usually via the manufacturer |
| Scope | Internal | Globally unique |
| Purpose | Run your operations | Identify the product externally |
| Format | Your choice | Standardised digits |
You need both. Marketplaces match your offer via the GTIN; your warehouse, orders and reporting run on the SKU. More on the identifier side in GTIN vs. EAN explained.
How granular should a SKU be?
The rule: one SKU per separately sellable, separately stockable unit. A t-shirt in three colours and five sizes isn't one SKU — it's fifteen. If two things can be ordered independently and counted independently, they need different SKUs.
Structuring a SKU scheme
A readable pattern beats a random number, because humans work with these daily. Something like BRAND-GROUP-MODEL-COLOUR-SIZE is easy to scan and sort. Three practical rules:
- Keep it stable. Never encode anything that changes — price, supplier, season, stock location.
- Keep it clean. Avoid spaces, umlauts, slashes and leading zeros; they break in CSV files, URLs and spreadsheets.
- Keep it consistent. The same logic across all product groups, or sorting and filtering stop working.
The mistakes that cause real problems
- Reusing a SKU for a different product after the old one is discontinued — historical orders and reports become wrong.
- Using the supplier's article number as your SKU. When you switch suppliers or two suppliers use the same number, the scheme collapses.
- One SKU for several variants. Stock becomes untrackable the moment one size sells out.
- Renaming at scale later. Every downstream system references the old value.
Why this becomes harder with many suppliers
For multi-brand retailers, the challenge isn't defining a scheme — it's applying it consistently to data arriving from dozens of suppliers, each with their own article numbers and variant logic. That mapping work is exactly what a PIM handles: supplier identifiers come in, your canonical SKU structure comes out, consistently on every import.