Objective: convert heterogeneous supplier feeds into a consistent product model that every webshop and marketplace understands.
1) Attribute mapping
- Source → Canonical: map vendor fields (e.g., name, desc, EAN) to a canonical schema with stable types.
- Defaults & fallbacks: apply sensible defaults when sources omit fields; keep provenance for audits.
- Normalization helpers: trim, lowercase enums, unify booleans, strip HTML where required.
2) Units & measures
- Unit registry: convert units to a base (e.g., cm, kg, L) with explicit precision and rounding rules.
- Dimension bundles: length/width/height and weight normalized for shipping and filters.
3) Taxonomy & categories
- Rules & lookups: category mapping by keywords, brand hints, and vendor categories.
- Attribute inheritance: category dictates required attributes (e.g., TV needs diagonal, resolution).
4) Identity & integrity
- SKU/EAN/GTIN checks: validate format, dedupe collisions, and protect canonical identity.
- Variant groups: relate color/size variants under one parent with coherent attributes.
5) Pricing & VAT
- Inclusive/exclusive: standardize price inputs; compute VAT-inclusive where needed.
- Currencies: normalize to base currency; store FX rate and timestamp.
6) Media & docs
- Asset hygiene: de-duplicate, validate mime/size, and enforce min resolution for hero images.
- Ordering & alt text: stable sort (hero → gallery) and meaningful alt text generation.
7) Validation & QA
- Schema validation: reject/flag invalid records with reasons and location.
- Quality gates: block publish if mandatory attributes missing for a category.
Outcome
A clean, searchable catalogue with predictable filters, accurate specs, and channel-ready payloads—no more ad-hoc fixes per destination.
What's next
Next up: building attribute taxonomies that scale—governing enums, ranges, and localized labels.