How to compress a PDF — what's inside a large file and what shrinks it

A ten-page contract should be 100 KB. When it arrives at 40 MB and bounces off an email limit, something inside is out of proportion, and it is almost always images. This guide explains what takes up space in a PDF and what "compressing" does to each part, so that the PDF compressor is used with the right expectations.

What makes a PDF large

A PDF is pages of drawing instructions plus the resources they reference — images, fonts, colour profiles, attachments — as How PDF files work: why some have text and some are pictures explains. Text and vector graphics are tiny: a page of prose is a few kilobytes. The bytes go to raster images (photos, logos, scanned pages), embedded fonts (a full Unicode font can be several megabytes), and incremental saves, where an editor appends changes to the end of the file without removing the old versions. A "compressed" PDF is one where those three have been dealt with.

Images: where the bytes are

Two things matter for an image inside a PDF: its pixel dimensions and its encoding. A phone photo placed in a Word document and exported keeps its 4000×3000 pixels even when shown at 5 cm wide — ten times the resolution any print needs (What DPI you actually need for printing). Compression downsamples images to a target density (150 ppi for screen and office print, 300 for quality print, 72 for review copies) and re-encodes them as JPEG at a chosen quality (How JPEG compression works — and what the quality slider really does). That alone typically cuts a bloated file by 80–95%. Scanned PDFs are the extreme case — every page is an image — and downsampling plus greyscale or black-and-white conversion is what makes them mailable; How to scan documents with your phone, and what 'scanning' actually does covers making them small at the source.

Fonts, revisions and junk

Embedded fonts should be subsetted — only the glyphs used — and most exporters do this; a PDF from a design tool sometimes embeds whole fonts, which a full rewrite fixes. Incremental saves and unused objects disappear when the file is rewritten from scratch, which is what a "save as" or a rebuild does. Metadata, thumbnails, attachments, form data and comments are small individually but worth stripping from a public copy for privacy as much as size (the metadata tool).

Choosing a compression level

• Light (200–300 ppi, JPEG quality 85): for print or anything with fine detail. Often halves the file with no visible change.
• Standard (150 ppi, quality 75): screen reading and office printing. The default for email attachments.
• Strong (72–100 ppi, quality 60): review copies and drafts. Photos soften visibly; text (real text) stays sharp because it isn't an image.

Text and vector content are untouched at any level; only images degrade. If a file barely shrinks, it had little image content to begin with — the honest outcome, and the compressor says so.

When a big PDF is right

A print-ready brochure, a portfolio of photographs, a scanned archive that must stay legible for a court: these are big because their images need to be, and compressing them is a mistake. Send them by link or file transfer instead. For everything else — the contract, the report, the form — compression is free, and keeping the original alongside costs nothing. PDF to JPG and the splitter handle the cases where only part of the file is needed.