At-a-Glance
- Deutsche Börse Group, a global financial market infrastructure provider, needed to govern data consistently across autonomous business entities – without creating centralized bottlenecks.
- By introducing a federated data governance model with a tiered architecture and custom roles, they enabled enterprise-wide metadata management that balances local agility with group-wide standards for consistency and context.
- Deutsche Börse’s approach enabled domain teams to document and classify data at business speed, reduced cross-team context hunting, and created a scalable data governance framework – proving that coordinated autonomy outperforms centralized control for complex and highly regulated organizations.
How do you govern data consistently across business entities – each with its own infrastructure, strategies, and operating rhythms – without stifling domain autonomy?
It’s a question that drives deep thinking in even the most data-centric companies – and Deutsche Börse Group was no different. But their answer had to combine autonomy with control.
So they built a model with nuance: a federated data governance framework that empowers domain-specific ownership without sacrificing high, company-wide standards.
Mira Boteva, Lead Product Manager for Metadata Management in one of the core business segments, and Shraddha Sharma, VP of Project Management for Data Governance on the Group level, joined Atlan Re:Govern to discuss why federated data governance beat out centralized control at Deutsche Börse – and what they learned along the way.
The Federation Imperative
Deutsche Börse Group is an international exchange organization and infrastructure provider, ensuring capital markets are fair, transparent, reliable, and stable through a range of products, services, and technologies. It houses several different business segments, which increases agility but also means that centralized data governance models risk becoming a bottleneck.
“Business segments have their own local data governance teams who are driving data governance for the business segment aligned with their own business strategies,” explained Shraddha. “Therefore, our data governance framework cannot be a central model, but rather federated.”
Taking Federated Data Governance From Principle to Practice
With such a diverse and dispersed business model, what does federated data governance look like at Deutsche Börse?
A Tiered Architecture
Their federated data governance model mirrors their organizational reality, with foundational pillars that build upon each other.
In practice, three groups take responsibility for different aspects of data governance:
A central data governance team sets the “what.”
This group sits at the company level, defining standards, principles, and guidance. An organization-wide policy serves as a single source of truth, ensuring business segments move in the same direction on governance fundamentals. This keeps expectations and enforcement consistent across the company, so teams don’t stray in different directions and throw standards out of sync.
Think of it as setting the rules of the road: data ownership must be clearly assigned, metadata management follows common patterns, security classifications work the same way across entities.
Local data governance teams own the “how.”
Individual business segments own the implementation. If the central team sets the rules of the road, the local teams are the drivers.
They decide how the standards set by the central team apply to their specific data domains, what their cataloging priorities are, and which use cases demand immediate attention. Local teams also define business glossaries that reflect their operational language, and implement data governance workflows that match their team’s rhythms.
The Data Governance Council and Data Stewardship Council bridge the gap.
These cross-functional groups align business units on compliance and standardization, without dictating local execution. To continue our driving analogy, the councils are the officials that enforce the rules and clean up the accidents.
Because the councils include stakeholders from different teams, they’re able to surface conflicts early (for instance, if two groups define “customer” differently), align on shared definitions, and pass along best practices.
As data environments become more distributed and workflows more complex, the councils’ responsibility for aligning meaning is critical for ensuring that all tooling has the semantic layers that reflect the business – including the information that lives in humans’ heads. This is key to closing the context gap (more on that later).
Balancing Agility with Consistency
Mira likened Deutsche Börse’s structure to a city with several buildings, each representing a business domain: “What we wanted to do is preserve the autonomy of each company when it comes to metadata governance, but at the same time, ensure consistency across the entire group.”
By treating data governance challenges as coordination problems – not control problems – the Deutsche Börse team refuses to make a binary choice between agility and consistency. Giving business units flexible frameworks while still ensuring enterprise-wide cohesion means the two can coexist.
And that flexibility is essential when trying to move fast enough to support teams that need rich context now – not after months of centralized, bottlenecked approval cycles.
Federated Data Governance + Context: Why Both Matter
While a federated data governance model made sense for Deutsche Börse’s business model, it’s more than just an organizational preference. As the race to maximize data utility intensifies – and regulatory scrutiny grows – it’s become a strategic asset.
Why Now?
It comes down to the unprecedented demand for data access, in real time and at massive scale. More users are discovering and working with data, but the tools they’re using – each of which offers its own semantic layer – struggle to put it all together. At the pace data is moving, teams and context become siloed. Add increasing regulatory pressure to the mix, and coordinating the two will be even more important.
In a federated model, context flows at the speed business units actually move. Local teams can document, classify, and govern their data domains without waiting for central approval on every field definition or access policy. That agility becomes critical when markets move fast and regulations like the GDPR demand demonstrable data governance controls.
Solving for context by design allows Deutsche Börse to keep moving and innovating quickly, without having to retrofit systems for new and evolving regulations. Managing data for the expanding regulatory environment is also critical for maintaining compliance across the company’s diverse business units, each of which may be subject to a different set of rules. Without context baked in, that task would be substantially more complex.
But when there’s a foundational plan for context, it gets captured where it’s created, by the people who understand it best and with guardrails to ensure it meets enterprise standards. That allows governance to scale with user demand, instead of becoming the friction point that stalls it.
The Path to Federated Data Governance
Beyond context, federated governance needs prioritization, automation, and a clear plan for metadata management in order to succeed. But those needs only become clear when teams start digging in, cataloging data, and capturing context at enterprise scale.
So when Deutsche Börse launched their Data Cataloging pilots, they deliberately chose two different data products:
- One ran on modern tech with clear ownership and better-documented metadata.
- The other lived in legacy systems where ownership was murky and data flows were opaque.
Both revealed the same fundamental truth: Cataloging has a knowledge-capture problem.
“Data cataloging is not an easy task. It needs human efforts and domain expertise,” Shraddha explained. “We needed to have business alignment and prioritization before we decided what goes into the catalog and for what purpose.”
This insight reshaped their entire approach. Simply pointing a tool at the data estate and expecting meaningful context to emerge would never work. Someone has to know what “customer” means in a specific domain. Someone has to document why a certain field exists and who uses it. Someone has to decide which datasets matter most for use cases or regulatory requirements.
In a federated model, that “someone” is distributed across business segments – which is both a strength and a challenge.
| The strength:Domain experts who actually understand the data are the ones documenting it. | The challenge:Those domain experts need the time, tools, and incentive to do this work, and to ensure their efforts tie to enterprise-wide standards. |
That led Deutsche Börse to three absolute needs that had to be in place for federated data governance to work.
Three Requirements for Data Governance and Data Cataloging
1. Strategic Prioritization over Comprehensive Coverage
The team at Deutsche Börse knew that trying to catalog everything at once would be futile. Instead, they started on the ground level, getting feedback from different teams and prioritizing the most critical information.
“We collected requirements across the group – which included the requirements from a business, IT, and data governance perspective – and we reached a level of 70+ requirements while aligning with four or five business segments in the group,” said Shraddha. “Then we compiled into key focuses, which included common data governance requirements like metadata, documentation, classification and lineage, and so on.”
Shraddha and Mira suggest asking business units for input like:
- Which data products enable our most valuable AI use cases?
- Which datasets carry the highest regulatory risk?
- Where are we currently blind to lineage or ownership?
Gathering input from different teams not only helps gain buy-in from the cross-functional stakeholders who will be impacted by any process changes, but also ensures that they can provide the right context that will make federated data governance work within their business unit.
2. Automation to Reduce Manual Burden
Through its data governance roll-out, the Deutsche Börse team realized that making adoption easy for all business units required automation. If every field definition required hand-coding, stewardship wouldn’t scale – and it definitely wouldn’t keep pace with demand.
Deutsche Börse is focusing heavily on automation – from metadata extraction to lineage mapping – to minimize the manual work required of domain experts.
Automation “continues to be a space where we are collaborating closely with Atlan to push the boundaries a bit further,” said Mira, acknowledging its role in data maturity and innovation.
3. Clear Processes for Metadata Management
Who approves business glossary terms? How do data contracts get created and updated? When does a local definition conflict need to be escalated to a group-wide council? These process questions proved just as important for Deutsche Börse as the technology ones.
Learnings from the Data Governance and Data Cataloging Pilots
Deutsche Börse’s pilots exposed a critical benefit: Uniting business and IT through shared documentation dramatically improved efficiency. Data teams stopped jumping between emails and Slack threads hunting for context. Business users could actually understand what data meant without reverse-engineering logic from code.
That’s what context looks like in action. When everyone is speaking the same language, the context gap closes, confusion and miscommunication decrease, and teams become more productive.
And in a federated model, it happens locally first, then connects to enterprise standards through governed processes.
Customization as a Feature, Not a Bug
Most enterprise software promises to work out of the box. But when the organizational structure is genuinely complex – multiple business entities, each with autonomous operations – that box doesn’t always fit the business.
For Deutsche Börse, this means software must be flexible enough to mirror how their organization actually works, not the other way around, going beyond standard configurations to federate data governance at scale.
“Customizing standard roles, into federated admin and data governance roles for a single domain that fit our governance structure, was the main challenge that we had during our journey,” recalled Mira.
But the effort was worthwhile: Deutsche Börse customized Atlan‘s existing features, introduced new ones, and created custom roles to support federated access. That allowed them to embed context in a way that fit the business, making data easier to understand, trust, and work with. As a result, users felt empowered to leverage data for decision-making, increasing adoption across business units.
Customizing Beyond Standard Configurations
A standard “admin” role in Atlan assumes centralized control: one team, full access, uniform policies. But Deutsche Börse needed federated admins who could govern their specific domain autonomously while respecting enterprise-wide guardrails. That required role hierarchies that didn’t exist in the standard configuration.
Similarly, they need governance roles scoped to individual business segments. For instance, a data steward in one entity had to be able to document and classify their data products without inadvertently affecting another segment’s catalog.
This level of customization isn’t a workaround – it’s essential alignment between governance model and organizational reality.
Four Lessons for Federated Data Governance
Throughout their journey, Mira, Shraddha, and their team honed in on important advice for peers building a federated data governance model that’s regulatory-ready:
1. Match the data governance model to the organizational structure
If the enterprise architecture is broad and composed of autonomous business units, a centralized data governance model may create more friction than value. Federation preserves local agility while enforcing enterprise standards – critical when systems and data consumers need context at the speed of business operations.
2. Treat cataloging as a strategic initiative, not a technical deployment
The “catalog everything” approach is set up to fail. Instead, start with business alignment:
- What problems are we solving?
- Which data enables business use cases or satisfies regulatory requirements?
- Who owns the domain expertise required to document it properly?
Human input and prioritization aren’t boxes to check – they’re essential features.
3. Choose strategic partnerships
Instead of simply buying a Catalog tool license, Deutsche Börse formed a strategic partnership with Atlan, focused on operational and cultural transformation. When governance requirements shift – and they will, especially as regulation and user demands evolve – having partner support in solving those problems will make adapting that much easier.
Look beyond the functional capabilities and at the support provided along the way. This helps ensure that implementation is fit for purpose, based on the organization’s needs.
4. Challenges validate the approach
If data governance were simple, federation wouldn’t be necessary. But challenges validate the federated model.
The lesson for other enterprises? If the organizational structure is complex, the governance tooling should be too – and customization is a feature that enables sustainable scale, not a bug to be avoided.
The Path Forward
Deutsche Börse’s story isn’t about achieving perfect data governance overnight. It’s about designing systems resilient enough to handle complexity, flexible enough to support both present and future demand, and rigorous enough to pass regulatory scrutiny.
“We are driving our data governance not only on the technical tool perspective,” said Shraddha. “Rather, on the perspective of operational and cultural transformation in the organization.”
From that vantage point, federated data governance is more than just an org chart decision – it’s a bet that distributed ownership, when properly coordinated, yields better outcomes than centralized control.
For enterprises caught between speed and regulatory rigor, that bet is looking to be one of the most reliable and sustainable paths forward.
What Can You Do with Federated Data Governance?
Atlan takes governance from passive to active, so you can turn data and AI into value – with speed, control, and precision. Book a demo to see how.
