YOUR DATA QUESTION
Expected columns and rows
Checks compare column names, values and row order against this fixture. They do not prove the query generalizes to every dataset.
SQL & DATABASE EXPLORER / ASK BETTER DATA QUESTIONS
Rows into answers.
One query at a time.
Explore a real practice database. Build queries, see the results and learn why joins and totals sometimes surprise you.
Loading the local SQL engine…
● ● ● your-next-question.sql
SELECT category,
COUNT(*) AS products
FROM products
GROUP BY category
ORDER BY category;Make the question precise.Then let the data answer.
0 / 30query challenges passed
Saved locally
One read-only statement per run. Every query starts from the same fictional database. Runs stop after 3 seconds; results are capped at 500 rows. Ctrl/⌘ + Enter runs your query.
Write a query, then inspect its result.
Explain the result
What does one output row represent? Which records were excluded? Would a different dataset change your answer?
KNOW THE SHAPE BEFORE THE QUERY
A small shop. Four connected tables.
All records are fictional. Prices and revenue are integer cents. Some customers have no orders; some products have no sales.
customers.id ← orders.customer_id
orders.id ← order_items.order_id
products.id ← order_items.product_id
orders.id ← order_items.order_id
products.id ← order_items.product_id
FOLLOW THE MATCHES
Why did my join add—or lose—rows?
This fixed visualizer joins customers to orders on matching customer IDs. It models the relationship independently of your editor query.
Read the grain
One output row represents a customer–order match, or an unmatched customer preserved by LEFT JOIN. A customer can appear several times. Counting rows is not the same as counting customers.
Take your database skills further.
Save and restore
Drafts, notes and completion checks save in this browser when storage is available. Query results are session-only. Export progress to move devices.
Your next steps
Download the 30-workshop handbook ↓
Download the practice database SQL ↓
These lessons use SQLite. PostgreSQL, MySQL and SQL Server differ in types, date functions and syntax. Production systems also require permissions, parameterized queries, indexing and backup practices beyond this read-only course.