Compress a PDF until it fits

What compression actually removes from a PDF

A PDF is a container. Inside it sit page descriptions, embedded fonts, metadata, and in most heavy files, images. When a document is too large, the images are almost always the reason. A ten page contract that has been scanned rather than exported is not ten pages of text, it is ten photographs of paper, and each one can carry several megabytes on its own.

Compression works on that in two very different ways, and knowing which one you are asking for saves a lot of confusion later.

Rewriting the structure

The gentle route leaves every page exactly as it is and rebuilds the file around it. Unused objects are dropped, repeated definitions are pooled, and the whole thing is written back out in a more efficient layout. Your text is still text. You can select it, search it, and a screen reader can still read it aloud.

The saving here is real but limited. On a file that was exported cleanly from Word or a design tool, you might see a few percent. On an old file that has been edited and re-saved many times over the years, the saving can be surprisingly large, because each round of edits left something behind.

Redrawing the pages

The aggressive route paints each page onto a canvas at a lower resolution, saves that canvas as a JPEG, and puts the JPEG back on a page of the original size. This is where the dramatic numbers come from. A 40 MB scan can land under 3 MB without becoming difficult to read.

The cost is that the page becomes a picture. The words are still visible, but they are no longer characters the computer understands. Search stops working, copy and paste stops working, and screen readers can no longer follow the text.

If the document has to stay searchable, choose the text safe setting and accept a smaller saving. If it only has to arrive under a limit and be readable, redrawing the pages will get you there.

Choosing the right setting

Four settings cover almost every situation. The resolution figures describe how finely each page is redrawn, measured in dots per inch.

Start with balanced. It clears the limits most people run into and stays comfortable to read on any screen. If the preview looks soft in places that matter, step back to light. If you are still over the limit, step forward to strong and check the preview carefully before saving.

Why your file was heavy in the first place

It was scanned rather than exported

Office scanners often default to 300 or 600 dpi in colour, even for a page of black text. Each page arrives as a full colour photograph. A twenty page document scanned that way can easily pass 50 MB, and none of that size is doing useful work if the pages are only going to be read on a screen.

Photos went in at full camera resolution

Dropping images straight from a phone into a document places them at their original size. A modern phone photo is several megabytes before it is placed on a page that will only ever display it at postcard size. The document carries the full file regardless.

It has been edited many times

Some tools append changes rather than rewriting the file. Delete an image, save, and the old version can remain in the file even though it no longer appears on any page. Files that have travelled between several people and several applications collect this kind of weight.

Fonts are fully embedded

To guarantee that a document looks the same everywhere, PDFs carry copies of the fonts they use. A handful of full font families adds up, particularly when only a few characters from each are actually needed.

The limits you are probably trying to clear

Most people compress a file because something refused to accept it. These are the ceilings that come up most often.

Two things catch people out. Attachment limits normally apply after encoding, which inflates a file by roughly a third in transit, so a file that sits just under the stated limit can still bounce. And a limit that says twenty five megabytes often means the whole message, not each attachment, so three files of ten megabytes will fail together even though each one passed on its own.

What compression cannot do

• It cannot open a protected file. A PDF with a password has to be unlocked first. Remove the password in the application that made it, then compress the open copy.
• It cannot bring back detail that was never captured. A blurry scan compresses to a smaller blurry scan. Rescanning at a sensible setting beats compressing a bad original.
• It cannot help much twice. A file that has already been compressed has little left to remove. A second pass mostly costs quality without buying size.
• It cannot restore text that has been redrawn. Once pages become images, the only way back to selectable text is optical character recognition, which is a separate job with its own error rate.

Checking the result before you send it

The preview shows the first page of the finished file, rendered from the actual output rather than from the original. That covers the most common problem, which is a document that has quietly gone too soft to read. It does not cover everything, so before sending anything that matters:

• Open the saved file and look at the pages with the smallest text, usually footnotes, figure captions and stamps.
• Check any page with a signature, a seal or a fine rule. Thin lines suffer first.
• If the document will be printed, look at it at one hundred percent zoom rather than fitted to the window.
• If you need search, press the find shortcut and look for a word you know is there. Nothing found means the pages were redrawn.

Doing it without handing over the document

Most compressors work by uploading your file, processing it on a server and sending back a link. That is fine for a holiday photo album and less comfortable for a signed contract, a medical letter or a passport scan. Your document sits on somebody else’s disk for as long as their retention policy says, and you have no way to verify it.

This tool does the work inside the browser tab instead. The page loads two open source libraries, and everything after that happens on your own machine. You can confirm it by opening your browser’s network tab before you start: the libraries come down, and your file never goes up. It also means the tool keeps working if your connection drops once the page has loaded, and that nothing needs deleting afterwards, because nothing was ever sent.