Optimize & secure guide
Make a PDF Smaller Without Losing Quality
Updated
Short answer: Use Compress PDF on Low: text and vector graphics are untouched, only oversized images are re-encoded at high JPEG quality, and the file structure is repacked losslessly.
“Without losing quality” means different things for text and for pictures. The text in a PDF can be made smaller with no loss at all. Photos can't shrink much without some loss, but it can be kept well below what you would ever notice.
Step by step
Step 1: Open Compress PDF and drop your file.
Open Compress PDF(runs on our server)Step 2: Choose Low (best quality). It keeps images up to 2,400 pixels on their longest side, at JPEG quality 82.
Step 3: Download the result and open it next to the original. Zoom to 200% on a photo and on the smallest text.
Step 4: If the saving is too small, try Medium and compare again. Stop at the first level where you can't see a difference.
Step 5: If the document will only ever be printed in black and white, converting it to grayscale removes colour data and can save a little more.
Open Grayscale PDF(runs on our server)
Why this works, and what to watch for
There are two kinds of saving. Lossless savings come from the file's structure: compressing streams that weren't compressed, packing small objects together into object streams, and dropping resources that nothing refers to any more. Compress PDF does all of this at every level, and none of it changes anything you can see.
Lossy savings come from images. A phone photo placed in a document is often 4,000 pixels wide even though it's printed 10 cm across. On Low, anything over 2,400 pixels is scaled down, which still leaves about 290 DPI across a full A4 or Letter width, more than most office printers resolve, and re-saved at quality 82, where JPEG artefacts are very hard to spot. An image is only replaced if the new version is at least 8% smaller, so images that were already efficient stay exactly as they were.
What isn't touched matters as much. Text stays text, with its embedded fonts, so it's identical at any zoom and still searchable. Vector charts and logos stay vectors. Images with transparency or masks are skipped, because JPEG has no transparency and would put boxes around them.
Two things compression can't fix. A scanned document is nothing but images, so it will always be bigger than one exported from Word; scanning at 200–300 DPI in grayscale instead of colour makes a far bigger difference than any compressor. And a PDF that is mostly text is already close to as small as it gets: what's left is largely embedded fonts, which have to stay for the text to display correctly everywhere.