What is the difference between ROW_NUMBER, DENSE_RANK, and RANK window functions?

ROW_NUMBER, DENSE_RANK, and RANK are all analytic (window) functions that assign a numeric value to each row within a partition of a result set. ROW_NUMBER() assigns a unique, consecutive integer to each row, starting at 1, with no gaps. It is often used for pagination or to give a deterministic order to rows. DENSE_RANK() assigns a rank to each row, but rows that tie on the ORDER BY expression receive the same rank, and the next rank is the next integer after the tie, with no gaps. For example, if the values are 10, 10, 9, the dense ranks would be 1, 1, 2. RANK() behaves similarly to DENSE_RANK() in that tied rows receive the same rank, but it introduces gaps: the next rank is the previous rank plus the number of tied rows. Using the same example, the ranks would be 1, 1, 3. In practice, use ROW_NUMBER() when you need a unique sequence, DENSE_RANK() when you want a compact ranking that ignores gaps, and RANK() when you need a ranking that reflects the true ordinal position with gaps for ties.