databricks

Zalando shares how they built a unified data foundation on Databricks to serve 50M+ customers across a complex microservices-driven data landscape. The approach centers on three pillars: (1) identity-based governance using Databricks Unity Catalog with a dual-catalog pattern and Dynamic Views to separate data creation from consumption; (2) a 'Metric as Code' semantic layer using Databricks Metric Views to centralize KPI definitions and eliminate metric divergence across dashboards, notebooks, and AI agents; and (3) conversational AI analytics via Databricks Genie, grounded in the governed semantic layer to deliver trustworthy, SQL-free insights to non-technical users like merchandisers and pricing analysts. The architecture reduces time-to-insight from hours to minutes for performance meetings and enables consistent metric definitions across all consumption surfaces.

How Zalando built a unified data foundation for AI and analytics on Databricks

The Foundation – Democratizing Governance with Unity Catalog

The Semantic Layer – Defining "The Truth" with Metric Views