Kimball or Data Vault? A decision you can actually make
They solve different problems. Use Data Vault to integrate and historise your sources; use Kimball to serve consumption. Most platforms need both.
SCIKIQ Academy08 Jul 2026 · 6 min readThe debate is usually framed as a religious war. It is not. Data Vault — hubs, links and satellites — is built for auditable integration and painless change: it never rewrites history, and adding a source means adding tables, not restructuring them.
Kimball dimensional models — facts and denormalised dimensions — are built for consumption: fast, predictable joins that non-experts can query.
On the lakehouse you rarely choose one. A common, pragmatic pattern puts Data Vault in the silver layer for integration and history, and dimensional models in gold for the business. Decide the grain first, use surrogate keys, and let each technique do the job it is good at.