DataHub supports compliance readiness with column-level lineage, named ownership, access policies, and a versioned history of every change, captured automatically across the estate.
It is encoded in the SQL that ran, but never captured anywhere an auditor can read. So one question means an engineer reverse-engineering a pipeline by hand, per field, per cycle. Most institutions can defend a fraction of the fields examiners ask about. The rest lives in tribal knowledge.
The cost: Remediation is measured in analyst-months, and it resets every reporting cycle because nothing was captured the first time.
Ownership was assigned once, in a spreadsheet, before two reorgs and a platform migration. The name on it has left. Nobody will attest to a definition they did not write, so accountability is the control that fails first, and it fails quietly until someone asks.
The cost: A governance control you cannot evidence is treated by an examiner as a control you do not have.
Pulling a current permissions list is straightforward. Proving who could see a PII column nine months ago, who approved it, and when it was revoked is a different question, and it is the one that gets asked. Access granted through a ticket queue leaves no durable record joined to the asset.
The cost: Point-in-time access evidence gets reconstructed under deadline, from systems that were never designed to retain it.
A regulation that applies continuously cannot be answered by a one-off exercise. The finding closes: a team documents the critical fields, tags the sensitive ones, fills the gaps. Then new pipelines ship, schemas change, people move, and the estate drifts back out of compliance, because nothing about the fix was continuous.
The cost: You pay for the same remediation more than once, and the next finding lands on ground you thought you'd already covered.
Demonstrate how a reported figure was computed: the exact source fields, the exact transformations, the exact aggregations, across core banking, trading platforms and loan origination, through the warehouse, into the report.
Lineage gets rebuilt by hand for the fields the examiner names, in the weeks before the response is due. It is accurate on the day it is written, unversioned, and cannot answer "what did the pipeline look like in Q3?"
Captured automatically as data moves, so the trace is a standing artefact rather than a project. DataHub publishes a mapping of these four lineage properties against the 14 principles.
Show that you know which data flows support critical functions, what they depend on, who is accountable for them, and what the blast radius of a failure or a third-party dependency actually is.
Architecture decks describe the estate as designed. The estate as running has drifted: new pipelines, new consumers, a BI layer nobody documented. So an impact question gets answered from a picture that no longer matches production.
Lineage is derived from what actually ran, across databases, transformation pipelines, BI tools and ML systems, so impact analysis runs against production reality and stays current without upkeep.
Answer which records hold a given individual's data, wherever it has propagated; show the lawful basis and the controls around it; and act on access, correction and erasure requests within a bounded window.
Someone classifies the source column correctly. Then it is copied into a staging table, joined into a mart, and pulled into a dashboard extract, and none of those inherit the tag. The scope of a DSAR becomes a guess.
A sensitivity tag applied to a column propagates through column-level lineage automatically, so downstream copies inherit the classification rather than escaping it, and the tag can sync back into the warehouse.
Maintain an accurate inventory of personal information by category, know which third parties receive it, and be able to delete or restrict on request, with a record that the request was actioned.
Privacy keeps a register; engineering ships weekly. The register describes the estate at the last review, and the gap between the two is exactly where an undisclosed data flow lives.
Assets are discovered by connector rather than declared by hand, classified with glossary terms and tags, grouped into domains and data products, and tracked for completeness against your own required properties.
A Type II report tests whether controls worked consistently across a period. For data, that means access was authorised and reviewed, changes were tracked, and confidential data was identified and protected, with sampling evidence to match.
Access approvals live in a ticket queue, reviews in a spreadsheet, classification in someone's head. Producing a sample means three exports and a reconciliation, and nothing joins the approval to the asset it was for.
Access requests carry a business justification, route to the right reviewer, and land in an audit log joined to the asset. Compliance Forms turn a review into a tracked task with completion analytics rather than an assertion.
An internal finding comes with a named owner, a remediation plan and a date the committee has been told. Closure requires evidence the gap was filled, and evidence it has not reopened since.
The steering deck says 60% complete. That number came from a project tracker maintained by hand, not from the estate, so it cannot be audited, and it stops being true the moment the programme's attention moves on.
Compliance Forms scope the assets, set the requirements, and assign them to the owners and stewards who actually know the answers, then report which assets meet or violate each requirement, continuously.
A regulated company without a deadline is not the audience for this page. If nothing is currently in force against you, the honest answer is that you have time to do this properly and shouldn't be rushed. If you have a finding, a date, or a regulation already applying to you, the rest of this page is about what you can put in front of the person asking.
Automated rather than assembled, which is what makes them survive between reporting cycles.
The artefact: a source-to-report trace naming the exact source fields, the exact transformations applied at each hop, and the exact aggregations that produced the final figure, across core systems, the warehouse and the reporting layer.
The artefact: a named accountable party per asset, split by ownership type (Technical Owner, Business Owner, Data Steward), with documented ownership at each step of the chain, not just at the source.
The artefact: an audit log of every access request in one place: the requestor, the entities selected, the business justification submitted, the reviewer's decision and comments, and the access duration granted. Reviewers are routed statically or dynamically to entity, domain or data product owners.
The artefact: a versioned history of metadata changes, so structure, ownership, definitions and classification can each be read as they stood at a past date, rather than as they stand now, which is the version an unversioned catalog is stuck with.
Checkout.com serves global brands including the Financial Times, GE, Sony and Patreon. As it onboarded more than 42 internal teams, manual governance stopped scaling, and masking sensitive PII across systems became the harder half of the problem. Using DataHub's Actions Framework, tagging a column as PII now emits an event to Kafka that propagates masking into Snowflake and notifies the user for confirmation. Temporary PII access auto-expires. Every metadata change is logged and replicated to Snowflake, which is what turned governance activity into audit-ready record.
"The most important bit of why we use DataHub's Actions Framework is to allow us to make real-time changes as soon as an event happens in DataHub."
A real regulated payments estate, not a single-warehouse demo.
Sources: Checkout.com customer story, the IDC Business Value Solution Brief on DataHub Cloud (March 2026), and the DataHub data governance product page.
Maps directly to the 14 principles of BCBS 239, covers embedding governance, ownership and accountability into daily operations, and makes the case for treating compliance investment as the foundation for what comes after rather than a sunk cost. Written against the reality that a decade after publication, most banks are still not fully compliant, because of legacy systems, fragmented governance and inconsistent data quality.
Most institutions can defend a fraction of the fields examiners ask about. The rest lives in tribal knowledge and manual effort that cannot scale. This guide covers why compliance initiatives fail to produce defensible documentation, the three impediments behind it, and the FinServ Compliance QuickStart: a pre-seeded FIBO-aligned glossary, a Federal Reserve MDRM ingestor, an opinionated configuration, and a CDE Data Steward Agent.
Share a few details about your goals, and we'll be in touch to schedule a personalised walkthrough tailored to your data environment, use cases, and roadmap.
We'll cover:
Let's get you the answers and insights you need.
A DataHub specialist reaches out within one business day. 30 minutes, no prep required.
Prefer to look around first?
Take the self-guided product tour