Lesson 1 of 88 min read
Decide what ETL testing must prove
Foundation 1: An ETL job extracts data from sources, transforms it and loads it into a warehouse that reports and decisions rely on. ETL testing proves that the right rows arrived, with the right values, exactly once, and that bad rows were handled visibly. Our synthetic nightly job reads seven order rows from a CSV, loads valid ones into fact_order and dim_customer in SQLite and records rejects with a reason. Dashboard design, source-system correctness and performance at scale are separate questions. Passing checks prove these rules on this file, not every rule in a real warehouse. Start with a short boundary note before adding tooling. This free orientation uses a synthetic orders file; it does not run Python inside the website.
Key takeaways
- 1Identify the source, target and mapping rule.
- 2Write an observable expectation and an excluded claim.
- 3Keep proposed deployment work distinct from performed checks.
Synthetic boundary note
Source: orders.csv (7 rows) Target: fact_order, dim_customer, etl_rejects Prove: completeness, correctness, uniqueness, visible rejects Out of scope: dashboards, scale
Go deeper
Expert notes beyond the basics
- 01
Reason through the boundary
An ETL job extracts data from sources, transforms it and loads it into a warehouse that reports and decisions rely on. ETL testing proves that the right rows arrived, with the right values, exactly once, and that bad rows were handled visibly. Our synthetic nightly job reads seven order rows from a CSV, loads valid ones into fact_order and dim_customer in SQLite and records rejects with a reason. Dashboard design, source-system correctness and performance at scale are separate questions. Passing checks prove these rules on this file, not every rule in a real warehouse.
- 02
Official reference
SQLite documentation: https://www.sqlite.org/docs.html . Use installed-version guidance and actual evidence, not assumed deployment results.
Common mistakes to avoid
- - Trusting matching counts without checking sums.
- - Pasting real customer rows into tickets.
Try it yourself
Write a rule, action, evidence and limitation for this topic.
Write your answer on paper or in a sheet first, then compare.
Interview questions
How to answer like a confident QA engineer
- Q1
What is the core promise ETL testing checks?
Model answer
Right rows, right values, exactly once, rejects visible
- Q2
What is outside this check?
Model answer
Functional correctness, real permission security, performance and deployed-version history need separate evidence.
Knowledge check
+5 points
What is the core promise ETL testing checks?
