How to Think About Schema Changes

Schema changes can significantly disrupt workflows, particularly when building data pipelines or performing traditional ETL processes. These changes may be caused by alterations to data sources or updates to business requirements. Fivetran addresses this issue in two ways: firstly, it automatically propagates schema changes from the source to the data warehouse, and secondly, it shifts the "transform" stage of ETL to the end of the process, allowing analysts to build data models using SQL without requiring engineers to reengineer pipelines. This approach helps reduce time and effort spent on handling upstream changes that may not affect analysis, enabling more focus on actual data analysis.