Compress PDF

About PDF compression

Where the bytes actually live

Compression isn't one thing — it's a stack of techniques applied to different parts of a PDF. In a typical 30 MB business report, roughly 60-90% of the file size is embedded images (photos, scanned pages, diagrams), 5-30% is fonts (a single typeface can appear 4-6 times across regular/bold/italic variants), and the remaining 5-20% is the actual page-drawing instructions and document structure. Knowing where the bytes live tells you where the savings are: cutting fonts and metadata won't help an image-heavy report, and cutting image resolution won't help a 100-page legal contract that's almost entirely text.

When compression is worth running

Three workflows dominate. Email attachments: most corporate mail servers reject anything over 25 MB, and many reject over 10. A compressed report sails through. Web hosting: page-load speed matters, and a 50 MB whitepaper that takes 30 seconds to download will be closed before it opens. Archival: storage is cheap, but full-text search and quick retrieval benefit when each document is a few MB rather than tens of MB. For most of these, image-heavy PDFs compress 30-70% with no visible quality change at normal viewing zoom.

The trade-offs

Compression that removes nothing perceptible at screen zoom (stream-level compression of content, font subsetting, removing orphaned objects) is genuinely lossless — text stays sharp at any zoom level. Image recompression and downsampling are visually lossless at typical viewing zoom but become noticeable when zoomed past 200% or printed at large formats. Two cases where compression is the wrong move: signed PDFs (any byte change invalidates the digital signature), and contracts heading to print where the signature block needs sharp 300-DPI rendering. For those, keep the original or compress before signing rather than after.

How browser-based compression compares

Server-based compressors (Smallpdf, iLovePDF, the Acrobat web service) upload your file, compress it on remote servers, and return a link. For a public marketing PDF that's fine; for an unredacted contract, NDA, or financial filing, it isn't. pdfmundo's compressor processes the file entirely in your browser tab — the PDF doesn't leave your device. Adobe Acrobat's desktop app offers more granular profile control and slightly tighter compression on edge cases. For most business documents, the gap is small and the privacy story matters more.

Common mistakes

Recompressing an already-compressed PDF is the most common one — every pass compounds artifacts, and after two or three rounds image quality degrades visibly. Always start from the original. Second: compressing a scanned PDF before OCR. A scan stored as 300-DPI images is 1-3 MB per page; after OCR (text layer + downsampled image) the same page can be 50-200 KB and searchable. Run OCR PDF first when the input is a scan. Third: compressing an encrypted PDF — the compressor cannot rewrite streams it cannot read. Remove the password in your PDF software first, then upload.

For the deeper engineering

This is the short version. For a full walk-through of PDF internals — what each compression technique does, why some PDFs resist compression, how to pick a quality profile for print vs. web vs. archival — see our engineer's guide to compressing PDFs without losing quality.