Understanding PDF File Size and Compression

Two twenty-page PDFs can differ in size by a factor of a thousand. Once you know what is inside them, the difference stops being mysterious and starts being predictable.

What occupies space in a PDF

• Text — negligible. A page of prose is a few kilobytes.
• Embedded fonts — a fixed cost of roughly 50 to 500 KB per font, paid once for the whole document.
• Vector graphics — small, and independent of how large they are displayed.
• Raster images — the dominant cost, and the only thing worth attacking in most files.
• Structural overhead, metadata and attachments — usually minor, occasionally surprising.

Doing the arithmetic

An A4 page scanned at 300 dpi is about 2480 by 3508 pixels — roughly 8.7 million of them. In 24-bit colour that is about 26 MB uncompressed per page. JPEG compression brings it down by a factor of ten or twenty, so a colour scan lands around 1 to 3 MB per page. Multiply by page count and the 90 MB scanned report stops being a surprise.

Lossless versus lossy

Lossless compression removes redundancy and reconstructs the original exactly. It is already applied to text and vector content inside every PDF, which is why re-compressing a text document achieves almost nothing. Lossy compression discards information permanently in exchange for much larger savings. Every meaningful reduction in a scanned PDF is lossy, and every one of them is irreversible.

Compressing an already-compressed image degrades it again without the corresponding saving. Always work from the highest-quality original you have.

Judging a result

The percentage saved is the wrong thing to look at first. A 95% reduction that renders the small print illegible is a failure; a 40% reduction that clears an email limit is a success. Judge by whether the document still does its job — read the smallest text, check the thin lines, look at any signature — and only then look at the number.

When to stop

If two successive attempts produce less than about 10% additional saving, the file has reached its practical floor. Continuing past that point trades visible quality for a rounding error. At that stage, either remove pages or change how you deliver the document.

Conclusion

Images dominate PDF size, lossy compression is the only lever that moves them meaningfully, and the honest test of a compressed file is whether it is still readable — not how impressive the percentage looks.