Glossary
Plain definitions for the vocabulary used across this site. Each term has its own anchor, so it can be linked to directly, and is marked up as a DefinedTerm inside a DefinedTermSet.
- SKUStock Keeping Unit
- The smallest uniquely orderable unit of a product. One SKU is one thing a customer can put in a cart at one price. Variants of size or color are usually separate SKUs grouped under a single product record.
- PIMProduct Information Management
- The system of record for product content and attributes. It governs how a product record is created, enriched, validated and approved before that data is published to commerce, warehouse and marketing systems.
- DAMDigital Asset Management
- The system of record for images, video and documents, with managed metadata, rights and versioning. A DAM is distinguished from a file server by the fact that assets carry structured metadata and are retrieved by query rather than by path.
- AttributeStructured product property
- A named, typed property of a product with a controlled set of possible values, such as Material = Stainless Steel. Attributes are what make a catalog machine-readable; free-text description is not.
- Attribute coverageCompleteness measure
- The percentage of products in a given category that have a value populated for a given attribute. Coverage measured by category is the fastest way to find where a catalog will fail customers, filters and feeds.
- TaxonomyClassification structure
- A hierarchical structure that assigns products to classes. A catalog typically needs more than one: a master structure for setup, a customer-facing structure for navigation, and a rollup for financial reporting.
- Boost and burySearch merchandising lever
- Rules that promote or demote specific products or attribute values in search results without changing the underlying relevance algorithm. Useful for margin, inventory and strategic placement, and dangerous when used to paper over bad data.
- Structured dataSchema.org markup
- Machine-readable annotations, usually JSON-LD, that describe what a page is about. It makes a page eligible for rich results in search. It must reflect what is visibly on the page.
- AEOAnswer Engine Optimization
- Work aimed at being cited by AI answer systems. Google treats this as ordinary search optimization and states there is no special markup for it; the differentiator is first-hand, non-commodity content.
- ETLExtract, Transform, Load
- The pipeline pattern for moving data between systems: pull it from a source, reshape it to the destination model, and load it. Most catalog integration work is ETL with validation added at the boundary.
- Controlled vocabularyGoverned value list
- A fixed, governed set of allowed values for an attribute, with a defined owner and a process for adding to it. Any attribute a customer will filter on needs one, because free text guarantees the same value will be spelled several ways.
- Rules engineDecision logic layer
- A layer that evaluates conditions across several attributes to decide an outcome: classification, workflow routing, a derived value, or a flag. Distinct from field-level validation, which only asks whether one value is allowed in one field.
- Defect reportingCatalog health monitoring
- Monitoring catalog health by writing one query per defect, where a defect is a single specific condition that should never be true of a product. Run on a schedule against a warehouse copy, ranked by impact, and routed to a named owner.
- Completeness scoreWeighted data coverage
- The weighted percentage of the attributes a product’s category and destination channel require that are actually populated. Unlike a pass/fail check it ranks the catalog, so enrichment effort can be aimed where it moves the most.
- Data regression testingDiscovery outcome testing
- Recording the products that should appear for defined search and browse scenarios, re-running those scenarios against production on a cadence, and alerting when results diverge. Catches discovery failures that break nothing and raise no error.
- Data warehouseAnalytical data store
- A queryable store holding copies of data from several operational systems, modeled for analysis rather than transactions. Where catalog health reporting belongs, because the useful questions span the PIM, commerce, inventory and behavioural data at once.