Compression shrinks images. It reduces their resolution and re-encodes them at lower quality. Everything else in a PDF is already small and already compressed.
So the honest expectation depends entirely on what is inside: a scanned document can lose 80 to 90 percent, and a text-only export from Word might lose ten. If a text PDF is huge, compression is the wrong tool and something else is going on.
Most advice about compressing PDFs skips the one fact that predicts the outcome. A PDF is not one kind of thing. It is a container, and what compression can do to it depends on what got put inside.
What is actually taking up the space
Open any large PDF and the weight is in one of two places. Either it is full of images, or it has an unusual amount of embedded font data. In practice it is nearly always images.
Text is extraordinarily cheap. A page of prose is a few kilobytes: some glyph codes, some positioning numbers, a reference to a font. A single full-page photograph at print resolution is several megabytes. One image page outweighs hundreds of text pages, which is why page count tells you almost nothing about file size.
Glyph codes and coordinates. Already compressed inside the file.
A photograph of a page. This is where the file size lives.
Downsampling and re-encoding images has an enormous amount to work with.
What compression actually does
Three things, in roughly this order of impact.
Downsampling: fewer pixels
A 600 DPI scan holds four times the pixels of a 300 DPI one and looks no better on a screen. Reducing resolution to match how the document will actually be used is the largest single saving available.
Re-encoding: lower JPEG quality
Each image is decoded and written back out at a lower quality setting. This is where visible artifacts come from, and it is the step to be careful with on pages containing text.
Housekeeping: removing the unused
Duplicate images referenced once each, embedded thumbnails, orphaned objects, unused font glyphs and metadata. Small individually, occasionally significant in files that have been edited many times.
Light, strong, and which to pick
The mode names describe how aggressive the downsampling and re-encoding are. The right choice follows from where the document is going, not from how big it currently is.
- Will be printedLight, 300 DPI floor
- Read on screen onlyStrong is usually fine
- Email attachmentStrong, then check the text
- Archival copyDo not compress at all
- Legal or medical recordLight, and keep the original
- Web downloadStrong, aim under 2MB
Telling when you have gone too far
Compression artifacts are easiest to spot in exactly the place people do not look. Zooming out makes a compressed page look fine. The damage is at actual size, on the smallest text.
JPEG compression is built for photographs, where gradual tonal change is the norm. Text is the opposite: hard black edges against white. Compressed too hard, those edges grow a faint grey halo and the letters lose their crispness. On small text this crosses from unattractive to unreadable quickly.
Why compressing twice makes things worse
Lossy compression discards information permanently. Running a second pass over an already compressed file does not recover any of it, and the second pass now works on the artifacts introduced by the first, treating them as real detail worth preserving.
The result is a file that is barely smaller and noticeably worse. If a compressed file is still too large, the right move is to go back to the original and compress harder in one pass, not to compress the output again.
How to get the best result
- Artifacts are introduced once, at a known level
- The compressor sees genuine image detail to work with
- Predictable quality across every page
- Easy to step back and retry at a gentler setting
- Original stays available as a clean starting point
Why it goes wrong
- Artifacts from the first pass get treated as detail
- Size barely improves, quality clearly degrades
- Damage stacks and cannot be undone
- Different pages degrade by different amounts
- No way back if the original was overwritten
What to do when compression is not the answer
Sometimes a file genuinely cannot reach the size you need while staying readable. That is not a failure of the compressor, it is a document carrying more information than the target allows.
Three alternatives are usually better than compressing into unreadability. Split the document and send it in parts, which is routine in legal work. Re-export from the original source with better settings, since a PDF generated badly from Word is often half the size when generated properly. Or remove what does not need to be there, starting with blank scanned pages, which are pure weight and cost nothing to lose.
Compress a PDF in your browser
Light or strong, with the file never leaving your machine. No upload, no queue, no sign-up.
Common questions
What people want to know before they trust the output.
01How much can compression shrink a PDF?
It depends entirely on what is inside. Scanned and image-heavy documents commonly drop by 80 to 90 percent because there is a great deal of redundant image data to remove. Text-only PDFs exported from a word processor might shrink by ten percent or less, because their content was already compressed when the file was written.
02Does compression reduce quality?
Image quality, yes. Text stays vector-sharp because it is not an image, unless the whole page is a scan, in which case the text is part of the picture and degrades with it. That is why scanned documents need more care than born-digital ones.
03Which mode should I use?
Light if the document will be printed or has to stay archival. Strong if it will only be read on a screen. When emailing, start with strong and check the smallest text at 100 percent zoom before sending, then step back if it looks soft.
04Why will my PDF not get any smaller?
Almost certainly because it contains no images to compress. A text-only document is already about as small as it can be. If it is nonetheless large, look for embedded font subsets, an invisible OCR text layer, or a high resolution image placed at small size on the page.
05Can I compress the same file twice to make it smaller?
You can, and you should not. The second pass works on artifacts left by the first and treats them as real detail, so the file gets meaningfully worse for very little size saving. Go back to the original and compress harder in one pass instead.
06Can I compress a password-protected PDF?
Not while it is encrypted, because the compressor cannot read the image data it needs to work on. Remove the password from a document you own, compress the unlocked copy, and re-apply protection afterwards if the file still needs it.