Consider the following preferences for a few groups of voters, with candidates labeled X to Z.
% of voters Their ranking
36% X > Y > Z
29% Y > X > Z
35% Z > Y > X
This is an instant runoff voting election. Y gets eliminated with the least first-choice votes and gets its ballots transferred to X. The winner would be X, with 65% of the vote. Because IRV satisfies the later-no-harm criterion, there is no way that either of the losing candidates (Y or Z) could be helped by the removal of a less preferred candidate.
For instance, the 35% of voters who preferred Z (last row, highlighted text) cannot possibly cause Z to win by removing X or Y from their rankings, since those rankings for Y will only be considered if Z is eliminated. And their rankings for X would only be considered if Y was then eliminated, and so on. In a nutshell, Z’s supporters didn’t hurt Z by also ranking Y and X further down the list.