Can you explain rank, dense rank, and row number?

In SQL window functions, RANK() assigns a unique rank to each row within a partition, but leaves gaps when there are ties. For example, if two rows tie for first, both get rank 1, and the next distinct value gets rank 3. DENSE_RANK() also ranks rows within a partition but does not leave gaps; the next distinct value after a tie gets the next consecutive rank. ROW_NUMBER() simply numbers rows sequentially within a partition, regardless of duplicates, so ties are arbitrarily ordered. Use RANK() when you need to preserve gaps for reporting, DENSE_RANK() when you want continuous ranking, and ROW_NUMBER() when you need a deterministic unique identifier for each row.