Collibra is a commercial data intelligence platform whose centre of gravity is governance: stewardship workflows, policies, and a business glossary, with a data catalog, lineage, quality and observability, a data marketplace, and AI governance built around them.
The cloud platform, or Collibra Platform Self-Hosted on infrastructure you choose. Metadata arrives through Edge, a site you run near the sources; lineage is harvested column-level from warehouses and BI tools; Data Quality and Observability is its own module with ML-generated adaptive rules.
How Collibra answers the questions Data Catalogs turns on.
| How it works | |
| Metadata model | A configurable operating model rather than a fixed schema, communities hold typed domains, which hold typed assets, with attribute types, relation types and complex relation types assigned per asset type. Metadata arrives through Edge, a site you run near the source |
| Lineage | Technical and business lineage as separate products for separate audiences, column-level, covering tables, columns, views and reports from Power BI, Tableau, Looker, MicroStrategy and SSRS. Stitching links harvested objects to catalog assets and leaves unstitched ones visibly grey. Built through Edge, the CLI lineage harvester reached end of life on 31 July 2026 |
| Search and discovery | Data Marketplace is the surface, search and filters across business domain, owner, system, classification and custom attributes; data products published with purpose, ownership and quality expectations; a basket to shop with; and an AI Copilot that takes questions in plain language |
| Business glossary | The product people buy the platform for, business terms in a configurable taxonomy of asset types, attributes and relations, moved through lifecycle statuses from onboarded to under review to approved by workflow, and linked to the technical assets they describe |
| Quality checks | A module of its own, Collibra Data Quality and Observability autogenerates adaptive SQL rules with ML, adds statistical and ML anomaly detection for the changes no rule would catch, and rolls results into scorecards. It overlaps the data-quality capability rather than deferring to it |
| Custom metadata | The operating model is the extension point, define your own asset, attribute, relation and complex relation types and assign them to one another, then build the processes around them in the Workflow Designer; a public marketplace distributes both, including access-request workflows and ServiceNow integrations |
| Connections | |
| Connectors | Catalog connectors Collibra provides, or your own JDBC driver, all reaching sources through Edge, with ingestion, profiling and sampling capabilities added per connection and a dedicated path for Snowflake. The documentation publishes no count and none is claimed here |
| Access | |
| Policy and compliance | The positioning rather than a feature, policies, business rules and data rules are asset types in the same operating model as everything else, carried through their lifecycle by a workflow engine with a Workflow Designer above it, so approval, ownership and evidence are the product rather than an add-on |
| Access requests | The strongest answer in this column, add assets to a data basket and checking out starts a request workflow: the owner of every element in it approves, technical stewards then provision the data, and the requester is notified when it lands, with the whole chain routable into Jira or ServiceNow |
| Cost | |
| Billing unit | Nothing published: collibra.com/pricing returns a 404 and every route ends in a sales conversation, so no unit can be recorded. For a commercial product that is a statement in itself, and it is worth reading beside the open-source rows here where the answer is zero |
Same headline facts as Collibra
vs Collibra: Managed · Serverless · Operational complexity: Low
vs Collibra: Open source (permissive) · Self-hosted · Free · Operational complexity: High · Java