Why a taxonomy? Without a consistent attribute system, filters break, search quality drops, and channels receive conflicting data. A scalable taxonomy governs names, types, allowed values and localization.
1) Enums that stay consistent
- Allow-list over free text: define canonical values (e.g., Color: Black, White, Navy).
- Synonyms & mapping: vendor inputs like “Midnight Blue” map to Navy via a synonym table.
- Versioning: enums evolve — add
valid_from/valid_toand deprecate with migration notes.
2) Numeric ranges & units
- Typed constraints: min/max, step, and required unit (e.g., TV diagonal ≥ 10", step 0.1).
- Validation: reject out-of-range values early; store the reason for QA.
3) Localization
- Labels & hints per locale: attribute and enum labels stored per language; fallbacks apply.
- Formatting: numbers, decimals and currencies follow locale rules.
4) Governance workflow
- Proposal → Review → Publish: changes to attributes/enums go through a lightweight approval with diff.
- Auditability: who changed what, when, and why — with impact analysis on categories.
5) Practical tips
- Keep attribute purpose-driven (search/filter/spec).
- Prefer short, human labels; keep machine names stable.
- Document examples and anti-examples right in the schema.
Outcome
With a governed taxonomy, your search facets stay predictable, feeds align across channels, and content teams move faster with fewer corrections.
What's next
Next up: Price & Stock Sync Strategies — latency budgets, FX handling, and conflict resolution.