Why Is My PDF So Big?

A PDF that's mostly text should be small, often under a hundred kilobytes. If yours is several megabytes or more, something specific is causing it, and it's almost always one of a short list of culprits. Here's what's actually going on, and what a compressor changes when you run one.

The usual suspects

High-resolution embedded images. This is the single biggest driver of PDF size by far. A single photo straight from a modern phone camera can be 4-8MB on its own. Include a few of those in one document and the file grows fast. Scanned documents are a specific version of this problem: a scanner set to a high DPI (dots per inch) essentially stores every page as a high-resolution photograph, even if the page is just black text on white paper.

Embedded fonts. PDFs commonly embed the entire font file used in the document, every character, every weight, every variant, even if your document only uses fifty of the several hundred characters in that font. A document using a couple of custom typefaces can carry several megabytes of font data that has nothing to do with your actual content.

Accumulated editing history. A PDF that's been edited and re-saved multiple times can accumulate redundant data from earlier versions that never gets cleaned up automatically.

What compression actually does

A PDF compressor isn't doing one thing. It's usually doing three, and they have very different trade-offs:

Downsampling images reduces the resolution of embedded images to something appropriate for how the PDF will actually be viewed. A photo scanned at 600 DPI for print doesn't need anywhere near that resolution if the PDF is only ever going to be viewed on a screen. Dropping it to 150 DPI is invisible to the eye at normal zoom, but because resolution is squared, cutting DPI by a factor of four cuts the actual pixel data by roughly sixteen times. This is where most of the size reduction comes from, and it's the one part of compression that involves a real, if practically invisible, quality trade-off.

Font subsetting rewrites the embedded font to include only the characters your document actually uses, instead of the whole typeface file. This is fully lossless. The document looks identical, there's just less unused font data along for the ride.

Object cleanup strips out redundant data left over from previous edits. Also fully lossless.

Why compressing twice doesn't help much

The first compression pass gets most of the available savings. Running a second pass on an already-compressed file typically only saves another few percent, because there's very little redundant data left to remove. The images are already downsampled, the fonts are already subset. If you need a smaller output than one compression pass gets you, adjusting the target resolution and starting over from the original file works better than repeatedly re-compressing an already-compressed one.

When not to compress

If your PDF needs to look right when printed, or contains a design/photography portfolio where image fidelity matters more than file size, aggressive compression can introduce visible softness when someone zooms in. Some files are simply large for a legitimate reason. Don't compress a print-ready file down to screen resolution and then send it to a printer.

Compressing

Compress PDF runs in your browser, and shows you the size before and after so you can judge the trade-off yourself before downloading.

Related tools

If the size problem is specifically about images inside a PDF you're about to create (rather than one you already have), converting your source images with Compress Image before building the PDF avoids the problem entirely.