TQ
Governance

Metadata with an owner, a definition, and a date you can trust

The catalog crawled the estate and came back with forty thousand tables, a third of them owned by people who left two reorganisations ago. It is complete, it is searchable, and nobody uses it to settle an argument. TopQuadrant governs the layer above the inventory: agreed definitions, named stewards, and a review cycle that keeps the record true.

Metadata · Governed business term
Business term Customer segment
Approved · v4

The marketing classification a customer is assigned to for targeting and reporting, reviewed each quarter by the segmentation council.

Steward D. Osei · Customer Data Attested 2026-07-02
Resolves to
Physical columnSystem
core.cust.SEG_CD Core banking Approved
crm.account.segment Salesforce Approved
warehouse.dim_customer.customer_segment Snowflake Suggested · 0.91
The third mapping was proposed automatically and waits on the steward. Nothing becomes authoritative without a person.

Cataloguing an asset is not the same as being accountable for it

A catalog answers what exists and where. It rarely answers which of these is the one to use, who says so, and when that was last checked. Those are governance questions, and they are the ones people actually arrive with. They are also the ones an agent arrives with: asked about revenue, it finds four fields called revenue, no signal about which is authoritative, and picks one.

200+

Prebuilt connectors, so the governed record stays in step with the catalogs, warehouses, and operational systems it describes.

Weeks

To a first governed domain in production, working from prebuilt models rather than a blank slate.

1

One governed definition per business term, resolving to the actual columns and systems that carry it.

Ownership and stewardship

Why does nobody trust the data catalog?

Because an entry with no owner and no review date is a search result, not a governed asset. Trust comes from accountability: a named steward per domain, a stated review cadence, and a visible date the record was last confirmed. Once an entry can go stale in public, it stops going stale.

  • Named steward per domain. Every asset resolves to a person, not a team mailbox or a former employee.
  • Attested, not assumed. Records carry when they were last confirmed and by whom, so staleness is visible rather than discovered.
  • Lifecycle states. Proposed, approved, deprecated, and retired are governed states with rules, not tags someone applies by hand.
CustomerD. Osei · Jul 2026ProductM. Whitfield · Jun 2026Financereview dueSupply chainno steward assignedGoverned domains · steward and last attestation

Glossary and technical metadata

How does a business term resolve to the actual columns?

The gap most estates never close is the one between what the business calls something and where it physically lives. Map each governed term to the columns, tables, and systems that carry it, and the mapping becomes an inspectable asset. Suggestions can be proposed automatically, but a steward approves before anything becomes authoritative.

  • Term to column mappings. A business definition connects to real physical assets, held as data you can query.
  • Suggested, then approved. Proposed mappings arrive scored, and a steward accepts or rejects rather than rubber-stamping.
  • Canonical by default. Wherever a column matching a governed term appears, it maps to the same definition, not a local copy.
Customersegment · D. OseiSEG_CDcore banking · approvedaccount.segmentCRM · approvedcustomer_segmentwarehouse · suggested~

Lineage and provenance

What makes lineage defensible rather than decorative?

A lineage diagram generated once is a picture. Lineage that holds up in a review is a set of governed facts: who asserted this relationship, when, on what basis, and what changed since. Held that way, it answers the question a regulator actually asks, which is not what the flow looks like but who signed it off.

  • Provenance on the relationship. Each link records its author, date, and basis, not just its existence.
  • Impact before change. See every downstream report, mapping, and integration affected before a definition is edited.
  • History that stays queryable. Previous versions remain available rather than being archived into a folder.
1AssertedD. Osei · Jul 022Basiscouncil minute3Impact9 reports4Retainedv3 queryableWho signed it offthe link is a governed fact, not a generated picture

Interoperability

Do we have to replace Collibra to do this?

No, and you should not want to. Catalogs are good at discovery and at breadth of coverage. What they lack is a governed source of meaning for the entries they hold. TopQuadrant sits alongside them as that source, and the catalog reflects the same definitions your stewards maintain rather than a second version of them.

  • Sync, don't compete. Collibra, Informatica, and Alation reflect the governed definitions rather than holding rival ones.
  • Open standards underneath. Everything is RDF, SKOS, OWL and SHACL, exportable in full at any time.
  • Readable by agents. The same governed record an analyst reads is what an agent resolves against, through the same interface.
CollibrareflectsInformaticareflectsAlationreflectsAgentsvia MCPOne definitionRDF · SKOS · OWL
What customers say
“Metadata is critical to be captured as semantics. It's not to live alone but to drive every key workflow this bank runs.”
Global 50 Bank
1
definition per business term across every catalog it appears in
100%
of governed assets carrying a named owner and a review date