English ▾
Free, Fast & Secure Online Tools
A three-page document has no business being 40 MB, and yet it happens constantly. The useful question is not “how do I compress this” but “what is in here” — because the right fix depends entirely on which of five things went wrong.
This is the most common cause by a wide margin. You place a photo from a modern phone into a document; it is 4000 pixels wide and several megabytes. The page displays it at 10 cm across, which needs maybe 1200 pixels at print quality — but the PDF stores all 4000.
Four such photos and you have a 30 MB file whose visible content would fit in 2 MB. The giveaway: a document with photographs where the file size roughly equals the sum of the original images.
A scanned document is not a document containing text — it is a document containing a photograph of text, one per page. Every page carries a full-page image, which is why a 20-page scan is far larger than a 20-page text PDF.
The test takes two seconds: try to select a word. If the cursor will not catch any text, every page is an image. Colour scans at 600 dpi are the extreme case; the same document scanned in greyscale at 300 dpi is often a tenth of the size and perfectly readable.
PDFs embed fonts so the document looks identical everywhere, which is the whole point of the format. But a full CJK font can be several megabytes on its own, and a document using four typefaces embeds four.
Subsetting — embedding only the characters actually used — solves most of this, and most modern exporters do it by default. If yours does not, a document with very little content and a surprisingly large file is the signature.
PDF supports incremental updates: when a file is edited and saved, the new data can be appended rather than the file rewritten. Do that twenty times and the file carries twenty layers, including content that is no longer visible.
This is why a PDF can grow every time you annotate it even though nothing was added visually. Exporting to a fresh PDF, rather than saving over the old one, discards the history.
A logo in a page header, placed separately on each of forty pages by a tool that does not recognise it is the same image, can be stored forty times. Well-built exporters deduplicate; not all do, and documents assembled by merging other documents are the usual offenders.
Before compressing, check what you are dealing with:
Inspecting the PDF’s internals will tell you how the page count, image count and fonts are distributed, which settles it quickly.
For images — the dominant cause — downsample before placing them, or compress the finished PDF, which re-encodes the images at a sensible resolution. This is the one fix that routinely cuts files by 80% or more.
For scans, rescan in greyscale at 300 dpi if you still have the paper. If not, compressing the PDF will re-encode the page images, which is the same operation.
For history and duplicates, re-export to a new file rather than saving in place.
One thing not to do: compress the same PDF repeatedly. Each pass re-encodes already-lossy images, and the text stays exactly the same size — so you lose quality without gaining much. Compress once, from the best version you have.