Page surgery, not content rescue
This workflow treats pages as objects to be oriented, trimmed, removed, and rearranged — it never touches what is written on them. That boundary matters at both ends. Documents that do not exist yet belong to the layout and generation workflows, which set type, margins, and page furniture from scratch. Scans that come in skewed, noisy, or damaged belong to the archival cleanup workflow, which straightens and denoises before anything else. And documents headed for signatures or distribution belong to the signing workflow afterward — page surgery changes every downstream byte, so it must finish before any seal is applied. What arrives here is a set of pages worth keeping that simply is not in printable shape.
Orient and trim first
Orientation is the foundation every later step stands on — a page count checked against a plan means little when half the pages are sideways. Rotate first, in quarter turns, until flipping through the document reads upright start to finish, then crop. The trim removes what the scanner added rather than what the document contains — the dark borders, the gutter shadow, the neighboring page's edge — and the result is page geometry clean enough that paper-size normalization and imposition have something consistent to work with. Checking rotation and trim together in one pass is cheaper than discovering a sideways appendix at the printer.
Fewer pages, chosen pages
Cutting is where page surgery goes wrong most often, and the safeguard is bureaucratic rather than technical — write the page plan down before editing. With a plan, the choice between deleting and extracting is simple arithmetic — if the survivors are the minority, extract them to a new document and leave the original untouched; if they are the majority, delete the unwanted pages in place. Both tools take page numbers, both act literally, and neither warns you that "page 12" in your head was "page 13" in the file — which is exactly why the plan is written, and why the output page count is checked against it at the end.
One paper size, one ink plan
Two properties decide how a printer treats the document, and both should be decided by you rather than the driver. Paper size first — a document mixing A4 and Letter pages comes out inconsistently scaled because the driver guesses per page, so every page is normalized to the destination standard before printing. Ink second — color earns its cost only where the content needs it; grayscale keeps photographic tonal range at a cheaper rate; high-contrast black and white sharpens text and line art and flattens everything else. A handout of prose wants the last, a product brochure wants the first, and the mixed case is usually grayscale.
Covers and imposition, then print once
The last moves give the document its final form. The unified cover goes on front — one prepended page that names the pile now assembled beneath it. Imposition is the other finishing move, and it is strictly for output that will be folded — booklet order rearranges pages so duplex printing plus one fold yields a correctly sequenced volume, with the cover sharing a sheet with the last page. Printing is the acceptance test for all of this — one test print, flipped through and folded, catches what every on-screen check misses, and a document that prints once, correctly, is the definition of done here.