A PDF often feels like a single sealed document, but inside it is a container holding text, fonts, vector drawings, and images, each stored in its own way. Compressing a PDF means shrinking those parts individually, and knowing which part is heavy tells you what any compressor is really doing.
What lives inside a PDF
A typical PDF holds a handful of ingredients. Text is stored as characters plus references to fonts, which is remarkably compact. Vector graphics — lines, shapes, and many logos — are stored as drawing instructions and are also small. Fonts may be embedded so the document looks identical everywhere. And then there are images, which are usually the largest single contributor, especially in anything built from scans or photographs.
Why images dominate
Because text and vectors are so efficient, the size of most PDFs is decided by their images. A report that is mostly words might be a few hundred kilobytes; add a dozen high-resolution photos and it balloons into megabytes. This is why a “compress PDF” action so often comes down to “re-compress the images,” and why a document with no images has little left to squeeze.
Lossless housekeeping
Some savings are lossless — they reduce size without changing how anything looks. Compressors can remove unused objects, discard duplicated data, compress the internal streams, and strip metadata or leftover elements from editing. On a bloated file this can help noticeably, and because nothing visible changes, it is the safe first pass. On an already-tidy file, though, these gains are modest.
Image downsampling and re-encoding
The big reductions come from working on the images. Downsampling lowers their resolution — sensible when a scan was captured at a far higher resolution than it will ever be viewed or printed at. Re-encoding then compresses each image more aggressively, often converting it to a more efficient format. Both are lossy: they trade image quality for size, which is why an over-compressed PDF shows soft or blocky pictures even though its text stays crisp.
Fonts and structure
Embedded fonts add size but guarantee the document renders the same everywhere, so removing them is usually a false economy. Some tools can subset fonts, keeping only the characters actually used, which trims size while preserving appearance. Structural cleanup — flattening layers or form fields that are no longer needed — can help in specific cases but rarely moves the needle the way images do.
Choosing a sensible level
The right amount of compression depends on how the PDF will be used. A document meant only for on-screen reading can tolerate lower image resolution than one destined for print. The practical approach is to compress, then check the images at the size they will actually be viewed. If text is your main content, expect small savings; if scans and photos fill the pages, expect large ones — and watch the images, since that is where any visible quality loss will appear.