Product data in baby & kids retail: safety data meets fast-changing sizes

Baby and kids is one of the more demanding retail categories for product data, and for a reason that isn't obvious from the outside: it stacks two hard problems on top of each other. Safety and labelling information matters more than in most categories, and at the same time the assortment is organised by age and size ranges that no two brands express the same way.

Problem one: age and size are a mess by default

The same item can arrive from three suppliers as “Größe 86”, “12–18 Monate” and “1Y”. Sometimes both age and size are in one free-text field. Sometimes they're in the product name. Occasionally the age is only implied by the category.

The fix is structural, and it's the same one that works in footwear and fashion: keep age range and size as separate, structured attributes, define your canonical values once, and map every supplier's notation onto them at import. Never let either live inside the product title — a customer filter can't read a title.

Problem two: compliance information carries more weight

Parents and regulators both expect more here. Age recommendations, warnings, material composition and manufacturer information aren't nice-to-have descriptive text — they're information the offer is expected to carry. The EU General Product Safety Regulation additionally requires certain operator and safety information in online offers; we cover the data implications in our GPSR overview. (Neither page is legal advice — check your specific obligations with qualified counsel.)

The practical consequence is the same as for any mandatory field: these have to exist per product, in the right language, on every channel — including for the brands that send you almost nothing.

Where it actually breaks: the longtail

Large brands in this category usually deliver reasonable data. The gaps appear elsewhere:

• Accessories — bottles, teethers, small textiles — often arrive with a name and a price.
• Small and seasonal brands deliver thin data with no attribute discipline.
• Fast-turning ranges mean the same gap-filling work returns every season.

That's the part that quietly consumes the team: the core range looks fine, and the last 40% of the assortment is where the mandatory fields are missing.

A workable structure

1. Canonical model per product group — age range, size, material, warnings, manufacturer information as defined attributes.
2. Supplier mapping stored once, applied to every future delivery, covering both field names and value notation.
3. Compliance values collected at supplier level where they're identical across a brand's articles — far less work than per product.
4. Completeness visible per group, so “what's missing” is a countable list rather than a worry.
5. AI enrichment for the gaps, with a review step before anything is applied — particularly useful for the longtail where research would otherwise be manual.
6. Channel-specific export, so each marketplace gets the fields it expects.

Frequently Asked Questions