Learn SQL. Work with data.

Ask a table for exactly the rows and columns you mean, and nothing else.

Collapse many rows into one answer per group, and know exactly which rows each number counted.

Put rows from two tables side by side, keep the ones that should survive, and find the ones with no partner.

Use one query inside another: as a value, as a list, as a table, as a named step, and as a set to compare against.

Put rows in the order the reader expects, break every tie on purpose, and hand back one page at a time.

Cut, clean, count, and combine text with a handful of functions that compose into anything.

Do arithmetic the way the database does it, and treat dates as values you can subtract, shift, and take apart.

Find what repeats, keep one of it, and summarise a column by its middle and its most common value.

Find what is missing from a sequence, what runs together, and what overlaps in time.

Compute a value across a set of rows without collapsing them, then rank, pick the top few, or compare each row to its group.

Let each row see the rows before and after it: running totals, changes from the previous row, gaps to the next, and moving windows.

Turn rows into columns and columns into rows, bucket values, fill blanks with zeros, and add the total line.

Walk trees of any depth, build paths, and generate the rows you do not have, with one recursive CTE shape.

Rates, active users, retention, funnels, and churn: the handful of query shapes behind almost every product metric.

Insert, update, and delete rows safely: from literal values, from other tables, with conditions, and inside a transaction.

Lesson with worked examples · No dedicated problem set