What is a data catalog?
A data catalog is a searchable index of the tables, files, reports, and data feeds an organization holds. It harvests technical metadata automatically from databases, the data lake or lakehouse, and business intelligence tools, including names, data types, refresh times, and lineage. People add the business context: a plain-language description, the owner, the sensitivity classification, and whether a dataset is certified for official reporting. The data dictionary, with its agreed field definitions, often lives inside the catalog.
Data catalog in practice
An analyst at a specialty retailer needs return rates by reason code. A search for “returns” finds four tables. The catalog shows that one is certified, owned by finance, refreshed nightly, and excludes online orders still in transit. Another is a store operations extract that stopped updating months ago. The analyst picks the right table without a round of emails. Fed to a text-to-SQL assistant, the same catalog entries tell the AI which table to query and what each column means.
What to watch for
- Assign owners before launch, since a catalog full of undescribed tables is soon abandoned.
- Mark certified datasets clearly, so users can tell official data from experiments.
- Check whether the catalog shows sample values, because previews of sensitive columns can expose data to people without access.