Why PDF files become so large

Almost every oversized PDF is oversized for one of four reasons, and knowing which one you have decides whether compression will help at all.

Scanned pages

This is the most common cause by a wide margin. A scanner does not produce text. It produces a photograph of a page, usually at 300 dots per inch, in full colour, one per page. A single A4 scan at that setting is somewhere between one and four megabytes before anything else is added. Twenty pages and you are at forty megabytes without a word of actual text in the file.

Compression works well here, because there is genuine redundancy to remove: the scan is colour when the original was black ink on white paper, and it is at print resolution when it will only ever be read on a screen.

Embedded images at full resolution

A document with photographs in it stores those photographs at whatever resolution they arrived at. A phone camera image is around four thousand pixels wide. Placed into a document at eight centimetres across, roughly ninety-five per cent of that detail is invisible and still stored.

Embedded fonts

Every font used in a document is usually embedded so it renders identically everywhere. A full font family with several weights can add several hundred kilobytes. This is worth keeping. It is also rarely the thing pushing a file over an email limit.

Accumulated revision history

PDFs can be saved incrementally, where each change appends to the file rather than rewriting it. Sign a document, add a comment, sign again, and all three states are still inside the file. This is why a document that has been round a review cycle is often several times the size of the original.

What actually reduces the size

If the file is scans or photographs, re-encoding the images at a lower resolution and quality will make a large difference. If the file is text, it will not. And any tool that claims otherwise is either doing nothing or turning your text into pictures, which makes the file larger and the text unsearchable.

Before compressing anything, open the document and scroll. If you can select the words with your cursor, it is a text document and compression has little to work with. If you cannot, it is a scan, and it will compress well.

Working out which cause you have

Open the file and try to select a word. If a text cursor appears, the document is digital and its weight is coming from images, fonts or revision history. If you get a selection box over the whole page, it is a scan and the pages themselves are the weight.

Then check the page count against the file size. A rough guide for a text document is under 100 KB a page; for a colour scan at 300 dpi, one to four megabytes a page. A twenty-page text document arriving at forty megabytes has something in it that does not belong.

The order to try things in

Start with the lossless steps, because they cost nothing.

Clearing metadata removes a few kilobytes and, more usefully, removes information you probably did not intend to send. Flattening a form bakes in the values and discards the interactive layer. Neither touches how the document looks.

If the file is still too large, the next question is whether it is a scan. If it is, compression will help enormously. If it is not, compression will make it worse, and the size is coming from revision history or embedded images at camera resolution.

Why saving a copy sometimes shrinks a file dramatically

Because it discards the incremental save history. Opening a document and doing a full save, rather than an incremental one, rewrites the file from its current state and throws away every superseded object.

This is why a contract that has been signed, commented on and re-signed can drop by eighty per cent when resaved with no visible change at all. If a file seems inexplicably heavy, this is worth trying before anything lossy.

A note on what compression is not

Compressing a PDF is not the same as zipping it. PDFs already compress their internal streams, so putting one in a ZIP archive typically saves a few per cent and adds a step for the recipient. It also gets stripped by some corporate mail systems.

If a file needs to be smaller, make the file smaller. Wrapping it in an archive is not that.

A rough triage table

| What you see | Likely cause | What helps | | --- | --- | --- | | Cannot select text | Scanned pages | Compress, and consider greyscale | | Text selects, few images | Revision history or fonts | Resave to discard history | | Text selects, several photos | Images at camera resolution | Compress, or replace the images | | Huge for its page count | Embedded attachments | Extract and remove them |

The triage matters because the wrong remedy is not neutral. Running an image compressor over a text document does not merely fail to help; it produces a larger file whose text can no longer be searched.