Most PDF size comes from images, not text
Text and vector content (fonts, lines, shapes) in a PDF are stored extremely compactly — a whole page of body text is often just a few kilobytes. The overwhelming majority of a large PDF's file size almost always comes from embedded images: scanned pages, photographs, or high-resolution screenshots pasted into the document.
How compression actually shrinks the file
A PDF compressor typically works on the embedded images in one or more of these ways:
- Reducing resolution (DPI) — a scanned page saved at 600 DPI has far more pixel data than most screens or printers can even use; reducing it to 150–200 DPI for on-screen viewing can cut the file size dramatically with no visible difference.
- Re-encoding with lossy compression (JPEG-style) — trades a small, usually imperceptible amount of image quality for a large reduction in size, by discarding image detail the human eye is least sensitive to.
- Removing redundant data — stripping duplicate embedded fonts, unused objects, and metadata that accumulate from repeated edits and re-saves.
When compression can't help much
A PDF that's already mostly text — a contract, an invoice, a report generated directly as a PDF rather than scanned — has very little image data to compress in the first place, so running it through a compressor won't meaningfully shrink it. Compression makes the biggest difference specifically on scanned documents and image-heavy files.
Try it yourself with the Compress PDF tool — everything happens in your browser, so the file never has to be uploaded anywhere to be shrunk.