Edit and compress
Change the file itself — its size, its markings, what it says about itself.
Adding to a page is safer than changing it
The tools here divide cleanly. Some draw something new onto a page — a watermark, page numbers, a signature. Others change the page itself, which is what compression and grayscale conversion do. The first kind is safe and reversible in practice, because the underlying content is untouched. The second kind is a trade you should make deliberately.
That divide also explains why editing existing PDF text is so difficult. The format pins every glyph to a coordinate, so touching one word disturbs the arrangement of all the others, and the typeface needed to redraw them may only be partly present in the file. Writing something new on top sidesteps every part of that, which is why the additive tools are dependable and the rewriting ones are not.
What compression actually costs
Reducing file size here works by rebuilding each page as an image. That's what makes large reductions possible, and it's also why the text stops being selectable afterwards. Nothing about that is hidden — the page says it, and the result tells you exactly how much came off.
That means a heavily compressed file and a searchable file are mutually exclusive here. If both matter, keep the original alongside the compressed copy and send whichever the situation calls for.
Before reaching for a quality slider
Most oversized PDFs are oversized because of images, not settings. A scanner defaulting to 600 DPI color for a black-and-white form produces a file on the order of twenty-five times larger than the job requires — nine times the pixels, three times the channels. Converting to grayscale often removes more weight than any compression level, and it does it without touching detail you need.
The other lever is page count. Check what you're working with, then drop the pages nobody asked for. Removing an appendix does more for file size than any setting, and unlike compression it costs nothing in quality.