Product Attribute Completeness for AI Shopping

Product attribute completeness is a coverage metric. The denominator is a declared set of fields, such as destination-required fields plus category-relevant attributes selected by the merchant. The numerator is the subset with values that meet the measurement's validity and provenance rules.

A basic product-level formula is valid populated fields divided by applicable fields. Catalog completeness can then report the distribution of product scores or aggregate field coverage. The method must state how it treats variants, conditional fields, empty strings, placeholders, inferred values, and products for which a field does not apply.

Completeness answers whether a usable value is present. Accuracy asks whether the value matches the merchant's source of truth. Consistency asks whether related values agree across the catalog, page, schema, and feed. Destination validation asks whether the record satisfies a platform's current rules.

These dimensions should not be collapsed into one claim. A product can have every field populated and still fail because an identifier is wrong, a price is stale, an image is unavailable, or a destination prohibits the category. A missing field can also be acceptable when the specification marks it conditional and the condition does not apply.

Paz evaluates product records against destination requirements and category-relevant attributes to identify concrete gaps. The result can support a catalog quality assessment and a product-specific proposal for a missing or weak field.

Generated values should remain proposals until they are supported by merchant data or approved through the configured workflow. Completing a field can improve the record's coverage and may resolve a validation issue, but it does not guarantee that an external platform will accept, display, rank, or recommend the product.