How PDF Compression Works (and Why Some PDFs Barely Shrink)

A PDF that is far too big to email usually surprises people, because the text inside it looks so simple. The reason is almost always the same: the file is mostly pictures. Understanding that one fact explains what a compressor can do, what it cannot do, and why some files barely change at all.

Why PDFs get big

A PDF is a container. Inside it sit text, fonts, vector drawings such as logos and charts, and images. Text and vector graphics are described with short instructions, so a page of plain text can be only a few kilobytes. Fonts add some weight when they are embedded, but rarely more than a few hundred kilobytes.

Images are a different story. A page scanned at 300 dpi is roughly 2,480 by 3,508 pixels, about 8.7 megapixels. Stored without any compression, at three bytes per pixel, that single page would be about 26 MB. Scanner apps and phone cameras store it as a JPEG instead, which brings it down to something like one to a few megabytes. Ten scanned pages can still reach tens of megabytes, which is why a short scanned contract can be bigger than a 300-page text book.

What a compressor can actually change

Compression tools for PDFs work on the images, because that is where the bytes are. There are three levers.

1. Re-encode the images at a lower JPEG quality. JPEG is lossy: it throws away fine detail the eye notices least. Lowering the quality setting makes the file smaller and the picture slightly less sharp. The saving is large at first and then flattens out, while the visible damage keeps growing.
2. Reduce the resolution of the images. Halving both the width and height of an image leaves a quarter of the pixels, so the saving is dramatic. The cost is that the page has less detail to show when you zoom in or print it.
3. Turn each page into a single picture. This is the most aggressive approach. It can shrink scans a great deal, but the text can no longer be selected, copied or searched, because the page is now just an image of text.

What a compressor cannot do is shrink text, fonts and vector graphics, which are already compact. That is why a PDF exported from a word processor often stays almost the same size.

Resolution or quality: which to lower first

For documents you will read on a screen, lowering the resolution usually looks better than lowering the quality. A page limited to about 1,500 pixels on its longest side works out to roughly 130 dpi on A4 paper. That is comfortable to read on a laptop or phone, though it is soft if you print it. Lowering quality too far tends to produce blotchy halos around letters, and that is harder on the eye than a slightly lower resolution.

If the document will be printed, keep more resolution (200 to 300 dpi is the usual range) and be gentler with quality. As a rule of thumb, scanned text that has to stay clearly legible should not go below a JPEG quality of about 60 to 70.

Why some PDFs barely shrink

If you compress a file and the result is only a few percent smaller, or the tool says it could not improve it, one of these is usually the reason.

• It is mostly text. There is little to remove.
• It was already optimised. A scanner app, a print driver or another compressor may have done the same work. Compressing a JPEG a second time gains little and costs quality.
• The images are an unusual kind. Simple tools only touch ordinary colour and greyscale images. Pictures that use transparency masks, or other special encodings, are left alone because rewriting them could damage the page.
• It is protected. Some encrypted or password-protected files cannot be rewritten at all.

When a compressor cannot make a file smaller, the honest result is the original file back. A tool that hands you a bigger file, or a damaged one, is a worse outcome than no change.

A sensible way to compress

1. Keep the original. Compression is not reversible, and you may need the sharp version later.
2. Pick a target, not a guess. If a portal says “under 1 MB”, aim for a little below it, as explained in our guide to meeting upload size limits.
3. Zoom to 100% and check the hard parts. Look at the smallest text, signatures, stamps and any numbers. These are the first things to turn fuzzy.
4. Back off if it looks wrong. Raise the quality or resolution one step and try again from the original, not from the compressed copy.
5. Consider splitting instead. If the limit is per file and quality matters, dividing a long document into two or three parts can be better than squeezing everything.

If you control the source

The cheapest saving happens before the PDF exists. For text documents, scanning in greyscale at 150 to 200 dpi is usually enough and produces files a fraction of the size of a 600 dpi colour scan. Many scanner and phone apps have a “small file” or “document” mode. Exporting from a word processor with a “minimum size” or “standard” option rather than “print quality” helps in the same way.

Compressing without uploading

Many online compressors send your file to a server. For a scanned ID, a contract or a medical letter, that is worth avoiding, and our guide on whether online file tools are safe shows how to check. The PDF compressor on this site runs in your browser: you choose a target size, it recompresses the images on your device and tells you the size before and after. If it cannot make the file smaller, it returns your original.