A scanned page can be 20 MB, and a scanned book is a problem

A text PDF of fifty pages is a couple of megabytes. A scanned PDF of fifty pages can be several hundred, because every page is a photograph. On a Mac used for paperwork, scans are frequently the largest category of documents by a wide margin.

Find them

find ~/Documents ~/Desktop ~/Downloads -name '*.pdf' -size +10M \
  -exec ls -lh {} + 2>/dev/null | sort -k5 -h | tail -20

Anything over about 10 MB is worth looking at. Open one and check whether you can select the text: if you cannot, it is an image and can almost certainly be made much smaller.

Reducing size without ruining it

1. Open the PDF in Preview.
2. File, Export, and choose the Reduce File Size Quartz filter.
3. Compare the result before replacing the original, since the built in filter is aggressive.
4. For scans you need to keep readable, scan again at 200 to 300 dpi in greyscale rather than colour.

The scanner settings matter more than any compression afterwards. Colour at 600 dpi for a page of black text produces a file ten times larger than greyscale at 300 dpi with no useful difference.

Text recognition changes the arithmetic

Running text recognition on a scan adds a searchable text layer without adding much size, and it means you can find the document later. It does not shrink the images, but it does make a smaller, greyscale scan much more useful, which changes what you are willing to compress.

Where scans accumulate

• Downloads, from every statement and form you ever saved.
• A scans folder the scanner software created and points at by default.
• Notes and Files, from phone scans that sync to the Mac.
• Mail, as attachments that are then copied into the downloads folder as well.

That last one is worth checking on the same pass, since Mail keeps its own copy of every attachment you opened, which means a large scan can exist three times on one Mac.