What is the medallion architecture (bronze/silver/gold)?
The medallion architecture is a layered model for refining data in a lakehouse that prepares data step by step from raw to analysis-ready. The three layers bronze (raw data), silver (cleansed and integrated) and gold (prepared for reporting and analysis) make the data flow traceable and reusable.
Also known as: bronze silver gold · multi-hop architecture · lakehouse layers
Where the medallion architecture is used
The medallion architecture structures the steps in which data is refined from source to finished report. Bronze holds the raw data as unchanged as possible (good for traceability and reprocessing). Silver cleanses, harmonizes and joins the data into a reliable, integrated layer. Gold provides the tables prepared for specific use cases, often dimensionally modeled, on which Power BI builds directly.
This approach is used above all in lakehouse environments such as Azure Databricks or Microsoft Fabric and combines well with data modeling approaches and governance.
A practical example
In the dy Project AG data platform, a large construction project worth over 1 billion CHF, data from SQL Server, Excel and REST APIs first landed unchanged in the bronze layer. In silver it was cleansed, unified and joined, and in gold it was then prepared into fast, validated tables for Power BI reporting. This kept it traceable at all times which figure was based on which raw data.
How it relates & how smiit uses it
The medallion architecture is a layering principle, not a data model. It defines in which stages data is refined, while approaches such as Kimball or Data Vault define how the tables within the layers are structured. It is a modern expression of ETL/ELT in the lakehouse and supports data governance because validated and unvalidated data are clearly separated. smiit uses the medallion architecture as a standard approach for lakehouse projects because it promotes traceability, reusability and clean responsibilities.
Common mistakes & misconceptions
- The medallion architecture is not a product but an organizational pattern that structures data into bronze, silver and gold layers from raw to refined.
- Many believe the three layers are a fixed rule. It is a guiding pattern that should be adapted to the use case rather than applied rigidly.
- A common error is to assume bronze already holds cleaned data. Bronze deliberately keeps raw data, while silver and gold clean and aggregate it.
Frequently asked questions
What do bronze, silver and gold mean?
Bronze is the raw data layer kept as unchanged as possible, silver is the cleansed and integrated layer, and gold is the layer prepared for reporting and analysis on which tools such as Power BI build directly.
Is the medallion architecture the same as ETL?
Not quite. The medallion architecture is a layering pattern for step-by-step refinement in the lakehouse. ETL or ELT describes the actual extraction, transformation and loading process that fills the layers.
Does it always have to be exactly three layers?
Three layers (bronze, silver, gold) are the common base pattern, but not a dogma. Depending on needs, intermediate stages can be added or, in simple cases, layers merged. What matters is the principle of traceable, step-by-step refinement.
What platform do you need for a medallion architecture?
It is typically implemented in a lakehouse, for example on Azure Databricks or Microsoft Fabric, often based on open table formats such as Delta Lake. The principle of staged refinement can, however, also be applied in classic data warehouse environments.
Related terms
Sources & further reading
Want to put this topic to work in your company?
Updated · Back to the glossary