What is Medallion Architecture?
A simple, scalable way to organize data from raw to analytics-ready.
The three layers: Bronze, Silver, Gold
Medallion architecture is a pattern that structures your data lake or warehouse into three layers, often called Bronze, Silver, and Gold. Data flows one way: from raw ingestion into Bronze, through cleaning and conforming in Silver, into business-ready datasets in Gold.
- Bronze: Raw data as landed from sources—minimal or no transformation. Preserves history and allows reprocessing.
- Silver: Cleaned, deduplicated, and conformed data. Often one table per entity, with consistent types and keys.
- Gold: Business-level aggregates, marts, and wide tables optimized for reporting, analytics, or ML.
Why use it?
Medallion gives teams a shared mental model: everyone knows where raw data lives (Bronze), where to find cleaned, conformed data (Silver), and where to consume analytics-ready data (Gold). It supports incremental processing, reprocessing when logic changes, and clear ownership. It’s technology-agnostic—you can implement it on Snowflake, Databricks, or other platforms—and works well with tools like Fivetran (Bronze), dbt (Silver and Gold), and medallion-style lakehouses.
Need help designing a Snowflake architecture?
Kundul helps teams design medallion-style Snowflake data engineering platforms, dbt model layers, and governed analytics foundations that scale cleanly.
Book a call