ELT (Extract, Load, Transform)
ELT is a modern data integration pattern where data is extracted from sources, loaded directly into a target system (usually a cloud data warehouse), and transformed in-place using the target system's compute.
Definition
ELT (Extract, Load, Transform) reverses the order of the last two steps in the traditional ETL workflow. Raw data is extracted from sources and immediately loaded into the destination system—often a cloud data warehouse like Snowflake, BigQuery, or Redshift. Transformation happens post-load, leveraging the target system's SQL engine and computing power. This approach is enabled by modern cloud warehouses' ability to handle semi-structured data and run complex transformations efficiently. ELT reduces data movement and processing overhead, making it faster and more cost-effective for large-scale data integration.
How It Works
1. Extract: Retrieve raw data from source systems. 2. Load: Stream or batch-load raw data directly into a cloud warehouse (minimal validation). 3. Transform: Run SQL queries, dbt models, or stored procedures in the target warehouse to shape data. 4. Query: Analysts and applications query the transformed data directly.
When to Use It
ELT is ideal for cloud-native environments, large volumes of data, and scenarios where schema evolution is common. Use ELT when your target is a cloud warehouse with strong compute and you want to minimize data movement. It's cost-effective for exploratory analytics and agile data teams. Avoid ELT if you need strict quality gates before data enters your warehouse or if you're using on-premises databases without strong transformation capabilities.
Relevant Tools
Last updated: Jun 17, 2026