
ETL Testing has come a long way from the early days of testers using sample data sets and writing code as second set of eyes and hoping to catch all defects; to understanding the need to identify all boundary test scenarios, setting up synthetic test data to simulate these scenarios for defect free testing. There are a lot of tools available, including homegrown and out-of-the-box SQL-based, that improve data quality and reduce testing time but don’t address the biggest challenges.
These tools either don’t support or are inefficient in 20% of situations that have transformations involved especially complex transformations. Even though these situations are less frequent, they are the biggest causes of unidentified defects.
So there is heavy reliance on skilled testers with or without tools to achieve good test quality, but the quality still varies from tester to tester and time to time because of human factors and experienced testers are expensive adding significantly to the testing costs
Missed test scenarios lead to gaps in test coverage:
Data setup is not automated:

