#05 The Zuma Roundtable Summary

ZUMA's latest Data Leadership Roundtable dug into one of the most ambitious shifts in modern data architecture: Data Mesh. The discussion went deep on implementation realities, cultural friction, and lessons learned from teams already navigating decentralised models.

Big thanks to this session’s contributors: Artur Yatsenko (Urban Sports Club), Kofi Nyamekye Manful (PERGOLUX), Kwaku Yeboah-Antwi (team.blue), and Suresh Bandela (Lucanet).

The ZUMAs: Joe Vaughan, Matt Brady

Data Mesh Isn’t a Template

Everyone agreed: there’s no one-size-fits-all. Data Mesh only makes sense when teams are mature enough to own data end-to-end. Some organisations are there, others are still managing fragile central pipelines and deeply embedded legacy systems.

Just because your architecture can support decentralisation doesn’t mean your organisation should decentralise. Readiness comes before rollout.

Autonomy, With Accountability

Balancing flexibility and standardisation was a core theme. Setting non-negotiables (for regulatory or operational integrity) while letting teams innovate within those guardrails is critical.

The group highlighted the role of contracts in making accountability concrete. If central teams are still firefighting accuracy issues, decentralisation hasn’t landed. Domain ownership isn’t symbolic — it should mean full responsibility for quality, testing, and documentation.

Semantic Layers Are Not Silver Bullets

Semantic layers came up often, not as a cure-all, but as a necessary layer for enabling data fluency. Participants stressed that clarity around definitions, clear ownership, and scheduled reviews are essential to prevent semantic layers from becoming shelfware.

There was a shared view that semantics should be tied to business questions, not just to alignment exercises. Common metrics are a foundation, but not the end goal. They’re only valuable if teams can use them to drive decision-making at speed.

Governance That Works

Top-down enforcement? Not so useful. Governance that’s co-created with stakeholders? Far more effective.

The group explored practical governance approaches, like councils made up of product, data, and business leaders who align on metric definitions and adoption. Tracking usage of shared KPIs was one tactic mentioned, but consensus pointed to aligning metrics with actual business goals instead of just investor expectations.

Governance is relational, not technical. It works best when there's trust, clarity, and visible value.

Change Management Is the Real Project

The technical transition is rarely the blocker. Change fatigue, unclear sponsorship, and cultural resistance are what derail Data Mesh efforts.

Migrating from legacy systems takes time. Leaders described multiyear transitions that required buy-in from execs and champions in each department. Success often hinged on the ability to connect infrastructure shifts to new business capabilities.

If you’re trying to mesh without executive sponsorship or local advocates, you’ll stall quickly.

When Data Mesh Makes Sense

Team maturity was repeatedly mentioned as the tipping point. Not company size, not stack complexity. Just the readiness of data producers to fully own quality and serve others through well-defined interfaces.

One participant summed it up well: if your central team is constantly blocking others, it’s time to explore decentralisation.

Advice for Germany’s Data Community

 👉 Start small, validate early (context matters more than hype) 👉 Define the non-negotiables (especially around regulatory risk) 👉 Assign definition owners and schedule regular reviews 👉 Build a culture of trust and clarity, not just autonomy 👉 Use investor metrics as input, not gospel

Final Thought: Data Mesh Is a Cultural Shift, Not a Config Change

Data Mesh isn't an architecture to be deployed. It's a cultural transformation grounded in trust, ownership, and the messy realities of collaboration. Without shared standards, clear accountability, and strong leadership, decentralisation just creates chaos.

When done right, though, it unlocks speed, resilience, and real alignment between data and the business.

🚀 Join the Conversation

Is your team pushing forward with a decentralised data model? What’s been your biggest hurdle: adoption, alignment, or autonomy?

We’re always looking to learn from data leaders building the next wave of infrastructure. If you're tackling this in Germany, let’s connect.

#ZUMA #DataMesh #DataGovernance #DataCulture