A common use case
Sharing just the summary section of a long report, pulling a single signed page out of a multi-page contract for a separate approval, or grabbing a handful of relevant pages from a reference document without sending the whole thing — extraction produces a new, smaller file containing only what you selected, in the order you set.
Combining non-consecutive ranges
Extraction is not limited to one continuous block of pages — adding more than one range (for example, pages 1-2 and separately pages 9-10) combines both into a single output file, in that order. This is useful when the pages you need are not next to each other in the source document, like a title page plus a specific appendix.
Extract vs. remove vs. split
If you are thinking "I want everything except a few pages," that is Remove Pages, not extraction — extraction is built around specifying what to keep, not what to drop. If you are thinking "I want to break this whole document into several separate files," that is Split, which is built for producing multiple output files rather than one focused one.
As a rule of thumb: extraction answers "give me just these pages," remove answers "give me everything but these pages," and split answers "cut this into pieces."