Explaining ETL, ELT and Reverse ETL

The aim of this page is to explain data warehousing concepts based on the particular example of ETL (Extract, Transform, Load) and Reverse ETL workflows. I dislike all the similar acronyms floating in the domain. ETL, ELT, CDP, CDI (Read https://evaparish.com/blog/how-i-edit as an anti-dose). Also this is really a

ETL (Extract, Transform, Load): The traditional data warehousing approach.

• The classical approach (https://www.ssp.sh/brain/etl/)
• Data is extracted from various sources.
• Data is transformed into a consistent format suitable for analysis.
• Transformed data is loaded into the data warehouse.
• For me, this is connected to old SAP R/3 and process chains I used to maintain when loading into SAP BW.

ELT

ELT (Extract, Load, Transform): A modern approach for Big Data.

• Data is extracted from source systems.
• Raw data is loaded directly into the data warehouse.
• Transformations are done on the data within the warehouse itself.
• Sounds like what DBT is doing

ELT

Reverse ETL (Reverse Extract, Transform, Load): Pushes data out to operational systems.

• As of 2023–2024, this is new and hot (see https://sourceoftruth.substack.com/p/why-martech-is-interesting-again)
• Data is extracted from the data warehouse.
• Extracted data is transformed to a format suitable for operational applications.
• Transformed data is loaded into various operational systems and applications.
• Typical destinations are CRMs like Hubspot and Salesforce, marketing tools like Braze and Marketo, finance tools like Netsuite, and many more.
• Census or Hightouch is a typical example or a Reverse ETL product

Reverse ETL