Reordering, deleting and rotating in one pass
Organising is the tool for documents that are nearly right. The pages exist and they are legible, but they are in the wrong sequence, or there are three blank pages from the scanner's feed, or the appendix ended up in front of the report. Rather than splitting the file apart and merging it back together, you rearrange the pages directly and export once.
This matters most for scanned documents, which arrive in whatever order the physical stack happened to be in. A duplex scanner that picks up two sheets at once, a stack loaded upside down, a page that jammed and was rescanned at the end — all produce a document whose page order does not match the document's logic.
Deleting pages properly
Removing a page here removes it from the output document entirely: its content is not carried into the new file, and it cannot be recovered from the result. That is the correct behaviour, and it is meaningfully different from the way some editors hide a page or mark it as not-for-display while leaving the data in the file.
The obvious candidates are blank separator pages, duplicate scans and the cover sheet from a fax. The less obvious ones are worth thinking about before you send anything sensitive: an internal distribution list, a page of pricing notes, a compliance annex meant for a different reader. Deleting is genuinely one of the cheapest privacy improvements available, because a page that is not in the file cannot be read out of it.
Work from a copy when the document is important. Page operations are quick and easy to get slightly wrong, and having the original to hand turns a mistake into an inconvenience rather than a loss. This matters most on documents you cannot simply regenerate — a signed contract, a scanned original, anything you received rather than produced.
Getting a long document into shape efficiently
On a document of any length, delete before you reorder. Removing the pages that should not be there first means you are shuffling a shorter list, and it prevents the common frustration of carefully positioning a page you then decide to remove.
Fix rotation last. Once the sequence is settled it is obvious which pages are the wrong way round, whereas rotating as you go tends to produce pages that are correct individually and inconsistent as a set.
Finally, check the finished order against the document's own numbering rather than against the thumbnails alone. Thumbnails at small sizes are surprisingly easy to misread, and a report whose pages run 1, 2, 4, 3, 5 looks entirely plausible until somebody reads it.
Organising versus splitting and merging
These three tools overlap, and choosing the right one saves a surprising amount of time. Organise when everything you need is already in one file and only the arrangement is wrong. Split when you want part of a document as its own file. Merge when you are bringing separate documents together. Reaching for the wrong one is how a two-minute job becomes a twenty-minute one.
The common inefficient pattern is splitting a document into individual pages, deleting the files you do not want, and merging the rest back together. That works, but it takes several passes, produces a folder of intermediate files you then have to clean up, and gives you more opportunities to lose track of which page was which. Doing it in one organising pass is faster and leaves nothing behind.
Where you genuinely do need to combine approaches is when pages from several documents must be interleaved — a contract with an annex from elsewhere inserted at a specific point, for instance. There, merge first so everything is in one file, then organise to put the pages in the order the document actually needs. Merging and then organising is almost always easier than trying to get the order right during the merge.