How would you design a lakehouse for a retail analytics scenario (sales, inventory, customers)?

• A. A snowflake schema with no governance.
• B. A data lake with no metadata management.
• C. A central data warehouse with a flat denormalized design and no governance.
• D. A central lakehouse for retail analytics with a star schema—dimensions for customers, products, stores; facts for sales and inventory—plus governance and programmatic access.

Answer: (D)

Designing a lakehouse for retail analytics hinges on combining scalable, governed storage with an analyst-friendly data model. A lakehouse gives you the best of both worlds: you keep the cost-effective, flexible data lake storage that can hold sales, inventory, customer interactions, and even unstructured data, while applying schema, governance, and ACID transactions so that analytics queries are reliable and performant.

Using a star schema is ideal for retail analytics because it centers analytic queries on clear, intuitive relationships. The dimensions—customers, products, and stores—provide the descriptive context you slice and dice by, while the facts—sales and inventory—record the measurable events and stock levels you want to analyze. This structure makes joins straightforward, aggregations fast, and BI tools or dashboards easier to use, which is essential for timely insights across sales, supply chain, and customer behavior.

Governance and programmatic access are also crucial. Governance ensures data quality, security, data lineage, and compliance, so analysts trust the numbers and data teams can manage access appropriately. Programmatic access—through APIs, notebooks, and data catalogs—enables automation, reproducibility, and scalable data consumption across teams and tools. Without governance and accessible APIs, the value of the analytics platform quickly degrades as data scales and usage grows.

In contrast, the other designs miss important capabilities: a snowflake schema omits the straightforward, BI-friendly querying of a star model and still leaves governance gaps; a data lake without metadata management becomes a data swamp with poor discoverability; and a central data warehouse with a flat, denormalized design and no governance lacks the flexibility and trust necessary for robust, scalable retail analytics.