SQL GROUP BY and Aggregation Practice — 147 Challenges (111 Free)

Aggregate functions over a whole table: 8 challenges (8 free)

No GROUP BY at all — COUNT(*), SUM(amount), AVG(price), MIN and MAX collapse every row that survived the WHERE into a single answer. Start with Counting Rows, then SUM, AVG, MIN, MAX: the two written for the syntax. Every one here is Easy and free, and three of them run on real public data — FDIC deposits, NYC assessed values, factory sensor readings. The cheat sheet has the whole list on one screen.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

Your first GROUP BY — one key, one table: 19 challenges (19 free)

One grouping key, one table, no HAVING: SELECT department, COUNT(*) FROM employees GROUP BY department. The rule that catches everyone is that every column in the SELECT list is either aggregated or named in the GROUP BY — SQLite will happily return a value from an arbitrary row instead of erroring. Start with GROUP BY Basics; Monthly Order Count and Day-of-Week Order Pattern (strftime) group by an expression rather than a column, which is the next step. The GROUP BY tutorial walks the mental model.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

Grouping by two or more keys: 32 challenges (20 free)

Add a key and the grain of the answer changes: GROUP BY country, category gives one row per pair, not one per country. This is where cross-tabs, cohort tables and per-group top-N live, and where a fanned-out join quietly doubles your SUM. Start with Counting Across a JOIN, then Survival Cross-Tab: Class × Sex; the Hard ones carry the pairs through CTEs and window functions.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

HAVING — filter the groups, not the rows: 30 challenges (21 free)

WHERE runs before the grouping and HAVING after it, so a condition on COUNT(*) or SUM(total) can only ever be a HAVING — and a condition on a plain column belongs in WHERE, where it throws rows away before they cost anything. Start with High-Volume Categories (HAVING), then GROUP BY + HAVING, both written for the clause. WHERE vs HAVING has the execution order in one diagram.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

COUNT(DISTINCT x) against COUNT(*): 12 challenges (7 free)

COUNT(*) counts rows, COUNT(x) counts the rows where x is not NULL, and COUNT(DISTINCT x) counts different values — after a join those are three different answers, and picking the wrong one is the most common wrong number in an analytics dashboard. Start with How Many Genres Do We Cover? (DISTINCT); Daily Active Customers and Customer Retention Cohort are the shape an interview asks for.

Counted from the challenge bank, September 2026. Easy first, then Medium, then Hard.

Aggregation is the skill the rest of the bank is built on, so it also turns up wherever the other topics do. If you want it with a join, a window frame or a CTE around it, the sibling pages list those: joins, window functions, CTEs, subqueries and CASE WHEN.