Structured data for products and variants

Product data should agree across the catalog, visible page and machine-readable descriptions. Variants make it easy to confuse a model's shared description with a particular size's price or availability. We begin by separating those responsibilities.

Distinguish the model from its variants

Google documents ProductGroup as a way to describe related product variants. We first establish shared attributes and those belonging to individual variants. Stable identifiers help preserve the relationship through updates.

Documentation: Google Search Central.

Define each value's source

Price, currency, availability and description should come from defined data. We avoid maintaining a second manually entered price list just for markup. The point at which inventory updates become visible is recorded too.

Example: a product in several sizes

The same model may have different availability by size. We test selecting each variant, entering through its URL and returning to the product. The visible offer and data for the selected variant should reflect the same state.

Check unusual cases too

Review includes missing images, withdrawn variants, promotions and availability changes. Valid syntax does not prove that the values match the offer. Validation results therefore need comparison with the actual product view.

Maintain it after launch

Catalog and template changes require another check of representative products. Reports distinguish data errors from search engine presentation decisions. Adding markup alone is not a promise of an enhanced result or increased sales.