PDF Compression Trade-offs: When 'Smaller' Means 'Worse'
A 90% smaller PDF is great until someone tries to print page 14 and the photo turns into mush. A short guide to picking the right compression level.
Every PDF compressor exposes some kind of slider — Low, Medium, High, or 72 / 150 / 300 DPI. The defaults are tuned for one thing: making the file smaller. They're not tuned for whatever you actually need to do with the result.
There are three places size comes from in a PDF, and you compress each differently.
First, raster images. These are usually the heaviest part of a PDF and the easiest to shrink. Compression downsamples them — a 4000-pixel-wide photo gets resampled to 1500 pixels and re-encoded as JPEG with quality 60. For email and screen viewing this is invisible. For print, it falls apart fast: at 150 DPI a half-page photo will look fine on screen and visibly soft on paper.
Second, embedded fonts. PDFs often embed entire font families even when only a handful of glyphs are used. Subsetting (which our compressor does automatically) keeps only the glyphs that actually appear in the document, often saving 100–500 KB per font.
Third, vector content and metadata. Drawings, form fields, JavaScript actions, and document metadata. These are usually small but can balloon if a PDF was generated by a chatty tool.
The practical rule: if the document is going to be read on screen and emailed around, aggressive compression is fine. If it might be printed or zoomed into, stay at 200 DPI or above for images. If it has signatures or form fields, test that they still work after compression — some tools strip interactivity to save bytes.
The other thing worth knowing: compressing the same PDF twice does not make it half the size. Once images are at quality 60, re-encoding them at quality 60 just degrades them further without saving more space. If a file isn't shrinking, it's already at its floor — find what's actually big inside (usually one or two oversized images) and target those.