Extract, Transform, Load (ETL)

Technical explanation

Extraction collects source data; transformation validates, cleans, joins, standardises, and applies business rules; loading writes the prepared result to a warehouse or other destination. ETL often uses a separate processing layer before the target receives data.

Business relevance

ETL supports controlled, repeatable preparation of data for reporting, migration, compliance, and operational use, especially when the target should receive only curated data.

Implementation example

A nightly process extracts CRM and billing records, standardises account identifiers, validates currencies, calculates revenue fields, and loads approved tables into a warehouse.

Limitations and common misconceptions

Pre-load transformation can increase latency and infrastructure complexity. Rigid pipelines may be difficult to adapt when source schemas or analytical needs change, and transformation errors can silently propagate without testing.

Topics

Sources