Definition
A product attribute schema is the definition layer of a catalog: for each product category, it declares which attributes exist, their data types and units, their allowed values, and whether each is required. It contains no product data of its own — it is the specification the product records are measured against. Crucially, a schema is defined per category, not globally: thread count belongs to bedding, wattage to appliances, and neither belongs to the other.
Key points
- Per-category by design. A single global attribute list forces nearly every field to be optional, because no field applies to every product.
- Each definition states five things: data type, unit, allowed values, how many values are permitted, and required or optional.
- It is what completeness is measured against — "80% complete" is meaningless without a schema saying complete against what.
- It is versioned. Flipping one attribute from optional to required changes the compliance status of every existing product in that category at once.
What does a schema definition actually contain?
For a bedding category the schema is a table of rows, one per attribute. Thread count: integer, no unit, single value, required. Material: single-select from a controlled list of cotton, linen, bamboo, and microfiber, required. Certifications: multi-select from a controlled list, any number of values, optional. Care instructions: free text, single value, optional. Nothing in that table names a product or holds a value. An appliances category would have an entirely different table — wattage, voltage, energy rating — with no overlap beyond the handful of universal fields such as brand and identifier that sit above every category.
Common pitfalls
- Maintaining one global attribute list instead of per-category schemas, so almost everything is optional and completeness scores stop meaning anything.
- Declaring an attribute a controlled selection without ever publishing the allowed-value list, so it degrades into free text within a quarter.
- Changing a required flag with no version bump and no backfill plan, which converts a schema edit into a silent catalog-wide compliance failure.
FAQ
How is a schema different from the product attributes themselves?
The schema is the set of definitions; the attributes as populated on a product are the instances of those definitions. The bedding schema says Thread count is a required integer with a single value. A product record says 400. Edit the schema and you change the rule for a whole category; edit the record and you change one product.
Should the schema be global or per-category?
Per-category, with a small shared core. A handful of attributes — brand, identifier, title — apply everywhere and belong above the category schemas. Everything else is category-specific. Merging them into one flat list is the most common way an attribute schema stops being useful.
Source
The idea of a published specification declaring which characteristics a product may carry and what value type each expects is exactly what the schema.org Product vocabulary provides at the vocabulary level. A category-level, required-versus-optional example of the same idea is Google's Merchant Center product data specification, which varies its required fields by product type.