Compress a PDF File
Two modes with very different outcomes: one repacks the file and keeps every pixel and character, the other turns pages into images and gets much smaller. The choice depends on whether you still need to select and search the text.
This only recovers structural overhead. Files from modern exporters often save nothing — when the result is not smaller, the original is returned unchanged.
Each page becomes a JPEG image: text cannot be selected or searched, and links, bookmarks, forms and annotations stop working. Right for scans, wrong for documents you still need to search or fill in.
Two different operations, one button
"Compress a PDF" conflates two unrelated things, and picking the wrong one wastes your time:
- Lossless optimization leaves page content alone and rewrites the container: packs scattered objects into object streams, removes redundant cross-reference tables, clears metadata. Quality and the text layer are unchanged, but the gain is a few percent at best and often zero.
- Scan compression renders every page to a JPEG and rebuilds the PDF from those images. Size drops to a third or less; the entire document stops being text you can select or search.
The test is simple: will you still need to search, copy or click inside this file? If yes, take lossless — even if it saves nothing. If no, trade the text layer for size.
Why lossless compression often saves nothing
A PDF's size is dominated by images and fonts, and neither can be touched in a lossless pass: re-encoding a JPEG is inherently lossy, and dropping font subsets turns text into garbage. What is left is structural overhead — cross-reference tables, uncompressed small objects, duplicated dictionaries. Word, Chrome and every modern reporting tool already handle those well, so optimizing their output again lands close to zero.
That is a deliberate trade here: if the rewritten file would be larger, keep the original and say why on the page. Plenty of tools skip that check and hand you a bigger file, leaving you to assume your document was the problem.
What scan mode actually does
Each page is drawn onto a canvas by the browser's rendering engine — with a white background painted first, otherwise JPEG encoding produces black — then encoded as JPEG at your chosen resolution and reassembled into a PDF that keeps the original page dimensions. It looks almost identical when opened, but internally the file has changed from vector pages into a stack of images.
That conversion is one-way: bookmarks, links, form fields, annotations and the searchable text layer all disappear. For scans, photocopies and brochures — documents that were images to begin with — nothing of value is lost. For contracts, invoices and reports that need to be searched, a lot is.
What this site does not do
Taking a 100 MB file down to 5 MB while keeping the text requires server-side Ghostscript or qpdf, doing image downsampling and font subsetting. That toolchain does not fit in a browser, so no such promise is made here. Extreme compression ratios require an upload — and not uploading is the whole point.