Ask a data governance team which critical data elements are missing an owner, and listen for the pause. Gartner reports that 57% of organizations don’t consider their data stewardship efforts successful. Yet the standards are written down, and the metadata needed to test them is already in the catalog. What’s missing is the check: something that continuously compares governance standards against reality without a steward having to verify each asset by hand.

What’s new: Collibra Control Tower

We are beyond excited to announce that Control Tower is now generally available. Control Tower continuously validates your Collibra Catalog across the Knowledge Graph and when a control fails, it alerts and routes the fix to the right user.

You can think of it as a governance enforcement engine. And it’s built with thousands of data governance users in mind.

You can easily build controls with the no-code interface in a few clicks or start from more than 100 pre-built controls across BCBS 239, AI governance, GDPR, Critical Data Elements and data products.

Controls run hourly, daily, weekly or on demand. Every run returns pass or fail for each control, sends an alert by email, Slack or Microsoft Teams when something fails and is kept in a run history. Users can drill-down into the failed assets with custom instructions to get things fixed right away.

How Control Tower helps

Governance leads write the standards. Stewards are expected to enforce them. In between sits a catalog full of data, named owners, linked terms, and dozens of other fields with no easy way to verify, at scale. The gaps stay hidden until something else exposes them: an audit, a supervisor’s request or a user who can’t find data they trust. Until now, closing that gap meant stewards checking assets by hand or engineers exporting metadata and maintaining SQL checks somewhere else. Control Tower automates those checks natively in Collibra, with a simple UI and alerts you only when a control fails.

Problems it solves

Coverage capped by headcount: manual review reaches whatever stewards have time for, and gaps hide in the assets nobody opened.
Checks only engineers can read: SQL on exported metadata carries no business context and changes only when an engineer has a free afternoon.
Reviewing everything to find a little: without exceptions, finding the 200 assets that matter means opening thousands.
Gaps discovered at audit time: a classification dropped in spring surfaces in the fall, when it is hardest to explain.

How Control Tower works

Let’s say you want to enforce a rule that says “every column in a Critical Data Element has a data quality score”.

You open the import menu and pick the closest pre-built control.

If nothing fits (spoiler alert: it does), you build it in the no-code query builder: choose the asset type, the pathway and the conditions.

In plain English: show me every critical data element missing data quality.

Controls can be easily built or imported from presets.

This pathway follows relations several hops out, so it checks that each element leads to a table, that table to a column and the column should have a data quality score. You test it once on demand to confirm it checks what your policy says, then schedule it daily.

The best part? When the next business unit adopts the same standard, they can just copy the control and point it at the domain where their Critical Data Elements live.

And if the control fails?

You don't open the 477 assets that pass.

You open one of the 23 that fail and drill down to fix it.

The alert has already landed in email, Slack or Teams with a convenient link to take action.

Alerts are sent to email, Slack or MS Teams

The steward doesn't need the dashboard to find it, either: the asset's own page lists every control it fails. When the steward assigns the owner in the catalog, the next run passes, and both runs stay in the history.

Assets display failed controls to easily fix them

And in just a few clicks, you are confident that your most critical assets are being watched and you have answers ready when leadership or the regulators ask.

Why you should be excited

Data governance leads: turn written standards into running controls themselves, from a pre-built pack or the no-code builder, without joining an engineering queue, then reuse them in the next domain instead of rebuilding.
Data stewards: start the day with a short list of assets that failed, each traced to the attribute at fault, or open any asset they own and see which controls it fails.
Risk and compliance officers: prove compliance with confidence, with a history that shows what was changed and when.
Heads of AI: check whether production AI follows internal and external policies. It ships with controls mapped to the EU AI Act and NIST.
Data consumers: check an asset's page for failed controls before building on it.

Use cases

AI Governance: Automate AI policy validation. For example, a control scoped to every AI use case checks whether any of its training data carries a PII classification.
Regulation-specific compliance: Map policies like BCBS 239 to live controls that monitor your data assets continuously. For example, Critical Data Elements must utilize data with the appropriate data classification.
Data products people can trust: Leverage Control Tower to monitor the Output Port to ensure the Data Product meets its service-level agreement (SLA) for data freshness, preventing users from using stale or outdated information.

Key takeaways about Control Tower

Control Tower turns governance standards into checks you can run. They stop living only in policy documents and start running against every asset in scope, on the schedule you choose. Failures surface when they happen, with the asset and attribute attached, and every run leaves a record. And because it runs on the knowledge graph you already built, you can start getting value today.

Manage by exception: stewards work from a short list of failures instead of sampling thousands of assets.
No engineering dependency: there are no export pipelines or SQL scripts to maintain; controls run where the metadata already lives.
One set of controls, two types of teams: the same controls serve regulated teams proving their standards and teams that just want a catalog people trust.
Maciej RusekMaciej RusekProduct ManagerCollibra

Maciej Rusek

Product Manager

Collibra