Finishing the SQL felt like winning… until the values take my good night sleep.

Whenever I built a data mart, my mind was laser-focused on one thing: ✅ Make the query work. Get the numbers out. Let the Datamart be shown.

Photo by Spencer Bergen on Unsplash

And when it finally ran with no errors, clean query, beautiful CTE. I’d sit back, proud and confident like, “Yep. This data is perfect. Nothing could go wrong.”

…except it did. 😅

Later I’d notice: some numbers didn’t add up, a column name wasn’t consistent, or the logic went slightly off.

That’s when I realized: building and testing require two completely different brain modes. And honestly? I just wanted to build, not play data detective all day. 🕵️‍♂️

Then I discovered dbt tests.

With just a few lines in YAML, dbt automatically checks your data for nulls, duplicates, relationships, or even your own custom rules.

Now, every time I run dbt build, it’s like having a silent QA engineer reviewing my work. It gives me peace of mind, knowing that my data isn’t just built… it’s trusted.

For anyone who lives in SQL, this feature alone makes dbt feel less like a tool and more like a teammate.