Metadata

Metadata is your own reference data on a product, stored as simple key/value pairs — for example SKU: ABC-123, Supplier: Acme, Season: SS26. It’s information you provide and control; Semantico never generates or changes it. Use it to carry the fields your business already tracks alongside each product.

A key point: metadata keys are shared across your whole workspace. The first time you use a key like Supplier on any product, that key becomes available across your entire catalog. So the same product data lives under a consistent, reusable set of keys rather than ad-hoc labels.

How metadata differs from everything else

Metadata is easy to confuse with other catalog concepts, so here’s the distinction:

Metadata — your own raw key/value data on a product. Never AI-generated.
Sections (custom attributes) — content Semantico generates with AI from a prompt.
Categories / Taxonomies — a controlled classification (a tree, or a reusable option list). See Categories and Lists and Taxonomies.
Lists — groupings of products for a workflow or export.

If you see the word “Attributes” in the dashboard, that’s just another name for your metadata keys — grouping attributes is a way to organize the same keys.

Setting metadata

There are several ways to add metadata, depending on scale:

On a single product — open a product and use the Metadata card to add key/value rows (Add Metadata).
In bulk via CSV import — any column in your CSV that isn’t a recognized product field (like sku, name, brand, category_path, product_lists…) is imported as a metadata key. This is the fastest way to load your existing data — name the columns the way you want the keys to appear.
Via the API — send a metadata object when creating or updating a product. See the Products API.

Managing keys

The dedicated Metadata page (in the dashboard nav) manages the keys themselves across your workspace — not individual product values. There you can:

Rename a key and set its display order.
See which products use a key.
Merge keys — consolidate several keys’ values into one. Useful for cleaning up duplicates like supplier and Supplier. Merging is permanent.
Delete a key.
Mark a key as locale-based when its value should differ per language (see below).

Locale-based metadata

Most metadata (a SKU, a supplier code) is the same in every language. For those, leave the key as a single shared value.

When a value genuinely differs per language, mark the key locale-based — then it stores a separate value for each locale, the same way translated content works. Importing a language-specific value for a key automatically turns it locale-based, so use this only when you really need per-language values.

What metadata is used for

Metadata isn’t just storage — it feeds several parts of the platform:

Filtering your catalog — filter products by a metadata key and value to find or segment products (for example, everything from a given supplier or season).
Exports — each key can appear as its own column in exports. Per key you can set:
- whether it’s included in the default export,
- the column header to use (its export label, falling back to the key name),
- and its order among your columns.
Publishing to your store — in a store integration’s field mapping, a metadata key can be the source for a store field (e.g. push Supplier into a Shopify metafield or a PrestaShop feature), and it can be pulled back from the store too.
AI prompts — a custom section’s prompt can reference a metadata value with a {{ metadata:KEY }} placeholder, so generated content can incorporate your own data. Note that AI does not read your metadata automatically — only the keys you explicitly reference in a prompt.

Recommended best practices

Standardize your key names. Keys are workspace-wide and matched exactly, so decide on a convention (casing, spelling) up front. Supplier and supplier are two different keys. Use the Metadata page’s merge to clean up any that drift apart.
Import in bulk with well-named columns. Since unknown CSV columns become keys, your CSV headers effectively define your metadata schema — name them deliberately.
Curate what you export. Turn off export for internal-only keys, and set clear export labels and order for the ones that go to partners or channels.
Keep values simple. Metadata values are stored as text — one value per key (per locale). It’s for reference data, not structured/nested objects.
Use locale-based keys sparingly — only when a value truly changes per language.
SKU is meaningful. If you set a SKU (or sku) metadata value on a product that has no SKU yet, Semantico uses it as the product’s SKU. Keep it consistent with your real SKUs.
Allow a moment for propagation. After a bulk import, newly added metadata can take a short while to appear in exports and filters while it’s processed.