The Medallion Architecture for data engineering is a powerful yet simple abstraction that drives architectural clarity. Per Strengholt (1),
“A Medallion Architecture is a data design pattern used to logically organize data in a lakehouse, using three layers for the data platform, with the goal of incrementally and progressively improving the structure and quality of the data as it flows through each layer of the data architecture”
At its core are three categories of data:
- Bronze tier – the home of raw ingested data
- Silver tier – the home of cleaned, filtered, and augmented data.
- I really like the use of the word conformed for the silver tier. A definition for conformed is “comply with rules, standards, or laws”. The data in the silver tier should map to a set of expectations for the data quality.
- Gold tier – the home of data that is ready for visualization, machine learning, or other purposes.
- This isn’t to say that you can’t use the Silver data for these purposes, but these applications are not the primary purpose of the silver tier.
This picture below, overlaid on the data engineering lifecycle (Using the data engineering lifecycle for data architecture – Lake Data Insights), provides a simple data flow for the medallion architecture.
I use the medallion architecture wherever I can in my solutions. To provide clarity, I name the storage areas in my data lake Bronze, Silver, and Gold. The scripts use the same convention. For example, the scripts that transform the Bronze data to the Silver data are located in a folder called BronzeToSilver.
While the concepts are simple, you can still find the devil in the details. The 350 pages of Peithein Strengholt’s recently released book (1) attest to this and provide some good guidance on nuanced topics. That said, keeping the fundamental intent of each layer in mind will help drive clarity into the detailed decisions.
The simplest approach that works wins, and the medallion architecture is often part of that winning formula.
(1), Building Medallion Architectures, Peithein Strengholt, 2025