Curating Catalog Records
Curating Glossaries and Catalog Records
A standard tells people what must be true about data, but it doesn't help them find the actual dataset sitting in some system and decide whether they can use it. That's the gap a catalog record fills. In Course 1 you nailed down a single trusted definition for one term; now you'll widen that habit. A catalog record is the plain-language label on a dataset that lets a colleague find it, understand it, trust it, and use it correctly without tracking you down. As a Data Governance Analyst, curating these records is some of the most visible, useful work you do, because it turns scattered data into something people can actually reach for.
What Minimum Metadata Actually Earns Its Place
The word that scares people here is "metadata," but it just means data about data: who owns this dataset, what it means, where it came from, what you're allowed to do with it. The trick is knowing which pieces a record genuinely needs. Pile on too much and nobody maintains it; leave too little and nobody can use it. The Minimum Viable Catalog Record names the smallest set that does the job: a dataset name, its business purpose, the domain it belongs to, an owner or steward, a plain definition, the critical elements inside it, a sensitivity classification, quality notes, a short lineage summary, the approved uses, and how often it's updated.
Every one of those fields exists to answer one of four questions a user is silently asking. Can I find it (name, purpose, domain)? Can I understand it (definition, critical elements)? Can I trust it (owner, quality notes, lineage, update cadence)? And can I use it responsibly (sensitivity, approved uses)? If a field doesn't help answer one of those, it's clutter. If a field is missing, one of those questions goes unanswered, and the user either gives up or guesses.
- Natalie: I found a dataset that looks perfect for the retention report, but I can't tell if I'm allowed to use it.
- Chris: What does the catalog entry say?
- Natalie: Just a name and the source system. No owner, no definition, no approved uses.
- Chris: Then you can't trust it yet. Without an owner to ask and a line on approved uses, you're guessing, and guessing is how people misuse data.
- Natalie: So the entry isn't really finished until someone could pick it up cold and know whether it fits.
- Chris: That's the whole job of a catalog record.
Notice that Natalie isn't blocked by bad data; she's blocked by missing metadata. The dataset might be perfect, but she can't tell.
