PDFs have a reputation for being bloated, but text is tiny. A hundred pages of pure text is smaller than a single photograph. When a PDF is enormous, the size is almost always coming from images.
Scans are the usual culprit
A scanner photographs each page at high resolution, often 600 dots per inch, and stores it losslessly. That is the right choice for archival work and wildly excessive for a document somebody will read on a laptop. Ten scanned pages can easily outweigh a hundred typed ones.
Embedded fonts and duplicates
Fonts add a little, especially when a document embeds several full families it barely uses. Merged documents sometimes carry the same image or font twice because each source file brought its own copy.
What compression can and cannot do
Compression re-encodes images at a sensible resolution and discards data nothing references. On a scanned document the savings are often dramatic. On a text-heavy file there is very little to remove, and you should expect the size to barely move.
- Mostly scans or photos → expect a large reduction
- Mostly text → expect very little, and that is normal
- Already compressed once → may not shrink at all
- Needs to survive printing at full size → compress cautiously