SQL lesson
GROUP BY and Report Grain
Practice GROUP BY and Report Grain with a worked query, common mistakes, and a checked next step. Create one summary row per requested dimension and predict output grain.
What you will learn
Create one summary row per requested dimension and predict output grain.
Before writing SQL, name what one result row should mean for GROUP BY and Report Grain.
Learning objectives and prerequisites
Each objective names an observable SQL behavior, with prerequisite topics linked in the order you need them.
Plan about 20 minutes for the explanation, worked example, prediction, and first hands-on attempt. Delayed review adds practice later.
- Objective: Create one summary row per requested dimension and predict output grain.
- Skill focus: group-by
- Prerequisite: review Aggregate Functions and Null Semantics before this lesson.
Example query and result
Study the compact worked query first, then open the editor when you are ready to change and run it.
- Starting rows: one fact row before grouping or joining.
- Result rows: one aggregate row per declared group.
SELECT customer_id, COUNT(*) AS order_count FROM orders GROUP BY customer_id ORDER BY customer_id;| Example output | Meaning |
|---|---|
| C001 | 3 | one aggregate row for customer C001 |
| C002 | 0 | zero-count behavior must be intentional when required |
Syntax pattern
Use the syntax pattern as a shape, not as a memorized answer. Replace table, column, condition, grouping, and ordering names according to the stated grain.
SELECT grain_key, COUNT(*) FROM fact_table GROUP BY grain_key;| Input grain | Output grain | Validation focus |
|---|---|---|
| one fact row before grouping or joining | one aggregate row per declared group | Rows, order, duplicates, nulls, and edge cases |
Common mistakes and why they fail
A common mistake is matching the visible rows while ignoring ordering, duplicate policy, null behavior, tie behavior, or the stated result grain.
Why it fails: GROUP BY and Report Grain checks changed examples, so a query that only copies visible rows can break when row counts, labels, nulls, or ties change.
PostgreSQL dialect notes stay explicit when syntax, date handling, transaction behavior, or comparison semantics matter.
Predict the result before you run it
Before opening the app, predict the output grain, the first column, and one edge case that could change the answer.
This is a reading prompt only; the app opens only when you choose to practice.
- Prediction: name one row that should appear or one row that should be excluded.
- Check: explain whether nulls, duplicates, ties, or missing relationships affect the result.
- Transfer: say what would change on a second dataset with different labels and counts.
Practice and related resources
Move to the app when you want the editor, checks, hints, and solution review.
Page highlights
- Clear task requirements
- Learning objectives and prerequisite links
- Original worked query and readable result example
- What each result row means
- Misconception and edge-case notes
- Predict the result before running the query
- Previous and next lesson navigation