Compress a PDF
without wrecking it.
The attachment bounces, the file is 34 MB, and the compressor you find turns your text into mush. Almost always the size is one thing: images at print resolution in a document nobody will print. Fix that and the rest takes care of itself.
Where the weight actually is
Open a large PDF and the culprit is nearly always embedded images. A scanner produces 300 or 600 DPI by default. A phone camera photograph dropped into a page is twelve megapixels. Text, by contrast, is tiny: a hundred pages of pure text is often under a megabyte.
Fonts come second. A fully embedded font set can add a few megabytes, and subsetting, which keeps only the glyphs the document uses, removes most of that without any visible change.
What to compress, in order
| Step | Typical saving | What you lose |
|---|---|---|
| Downsample images to 150 DPI | 60 to 85% | Nothing on screen, detail in print |
| Re-encode images as JPEG q80 | 20 to 40% | Slight softening on photos |
| Subset fonts | 1 to 3 MB | Nothing |
| Strip metadata and thumbnails | small | Nothing you will miss |
| Flatten to images | varies, often larger | The text layer. Avoid. |
Do them in that order and stop as soon as you are under the cap. The first step alone usually gets there.
Choosing a resolution
For a document that will only be read on screen, 150 DPI is more than enough. At typical viewing sizes it is indistinguishable from 300. For something that might be printed on an office printer, 200 to 220 is a safe floor. Only go to 300 if it is going to a commercial printer.
Scanned text is the exception. Type survives downsampling worse than photographs because the eye is sensitive to edge sharpness on letterforms. Keep scanned pages at 200 DPI or higher, and prefer a lossless setting for pure black and white scans.
Never compress the same PDF twice. JPEG re-encoding is lossy every pass, and the artefacts compound quickly into visible blocking around text.
Hitting a hard cap
If a 10 MB limit is the target and downsampling has not got you there, in order: split the document and send it in parts, remove pages nobody needs, or put it somewhere and send a link. Splitting is almost always better than compressing further, because the second pass is where quality falls off a cliff.
What flattening costs
Some tools compress by rendering every page to an image and rebuilding the PDF around it. The file may get smaller, and often does not. What certainly happens is that the text stops being text: it cannot be selected, searched, copied or read by a screen reader, and the document becomes inaccessible.
For a document going to another person, that is a bad trade at any file size.
Privacy
The documents people most want to shrink are contracts, invoices, medical letters and identity papers. Those are exactly the documents that should not be uploaded to a free web service in exchange for a smaller file. A compressor that works in the page does not send anything.