Tiers of one dataset by quality level, from raw landing zone to archive.
1 medallion architecture example
Five-tier medallion architecture
Raw, anonymised, staging, aggregated and archive tiers of one survey dataset.
Each tier is the same data at a different quality level, and each card says what it holds, who writes it and how. The aggregated tier is focal because that is what the business reads, and the archive promotion is dashed.
A medallion architecture is a data design pattern, popularized by Databricks, that organizes a lakehouse into tiers of increasing quality: bronze holds raw ingested data, silver holds cleaned and conformed data, and gold holds business-ready aggregates. Many teams add a landing zone before bronze and an archive tier at the end.
Read it left to right. Each tier takes data from the one before, applies validation or transformation, and serves a different audience.
When to use a medallion architecture diagram
Use it to explain a lakehouse design to analysts and stakeholders who need to know which tables to trust.
Use it to define data quality rules and ownership per tier.
If you need to show every source system and protocol, a data platform integration diagram gives more detail.
How to make a good medallion architecture diagram
Name each tier and write in one line what happens to data on the way in.
Show example datasets in each tier so the abstraction becomes concrete.
Note retention and access rules per tier, since they usually differ.
Keep the color convention consistent: bronze, silver, gold.
