Make a PDF smaller

Find the reason the file is big

A PDF is big for one of three reasons. It holds scanned page images, so every page is a photo. It holds photos placed at camera resolution nobody needs. Or it holds embedded fonts that never got cleaned up.

Open the file and look at one page. If you can select the text with your cursor, the pages are not images, and the weight sits in the pictures. If you cannot select anything, the page is a scan, and the scan is most of the file size. Size breakdown shows the split per page when guessing feels risky.

Each reason has its own lever in the compress tool. Placed photos get re-encoded at 150 dpi and quality 75. Whole scanned pages go through the guarded raster reserve, which renders page after page inside a fixed 60 to 92 quality range. Embedded fonts get subsetted to the glyphs the pages draw. The verdict names the lever that won, so the result teaches you what your file was made of.

Pick by the size you need

Some jobs have a hard number attached: a form that rejects anything over a stated limit, a mail gateway with its own ceiling, a portal that names a limit in KB. If you have a number, use Compress PDF and type it in. The number is a ceiling the tool aims under: it returns a file that fits, or the smallest honest file with the exact distance left to go.

Mind the two ways of counting. This tool counts 1,024 bytes to the kilobyte and 1,024 KB to the megabyte, so 100 KB here is 102,400 bytes and 1 MB is 1,048,576 bytes. A counter using flat thousands stops earlier - set 97 KB instead of 100, or 976 KB instead of 1 MB, and the dispute disappears. The exact-size guide walks through the whole trap.

If you have no number and just want the file lighter, leave the target box empty: the tool falls back to 500 KB, or just under your file size when the file itself is smaller. A file already under the target returns unchanged. Stop when the file is small enough to send - and when it matters, go by the size the receiving system reports, since counters disagree about kilobytes.

Do it in four moves: open the compress tool, drop the file in, type the size you need, and download the result the verdict points at. It runs in this tab; with no target typed, the box falls back to 500 KB.

What compression cannot do

Nothing here rewrites the words on the page. When the raster reserve redraws a page as an image, it writes the page's own text back first as an invisible layer, so selection, search, copy and paste keep working. That layer is why the tool refuses any squeezed candidate missing over 2% of its non-space characters.

If a file is already small and dense, there may be nothing left to take out. An already-small text file with no images has almost no slack in it, and no measured floor exists here for how little that can be - a tool that promises a fixed percentage on every file is guessing.

The same honesty applies to quality. The verdict names the lever that won: if an image method won, pictures were re-encoded at 150 dpi and quality 75, and the raster reserve works inside 60 to 92. Either way, compare the preview before sending. The quality guide shows how to read that trade before you pay it.