Why Your Scanned PDF Is 40MB, and How to Fix It

What is actually taking up the space

A scanned PDF contains no text. It contains photographs of pages, and the size of a photograph is driven by three settings: resolution, colour depth, and compression.

Resolution dominates. A page scanned at 600 DPI contains four times as much data as the same page at 300 DPI, because doubling the resolution quadruples the pixel count. Many scanners default to 600 for no better reason than that it sounds thorough.

Colour depth is next. Scanning a black-and-white letter in full colour stores three colour channels for a page that has one meaningful tone. Greyscale roughly thirds the data with no visible loss on a text document.

Compression is the third. Some scanners save pages as effectively uncompressed images inside the PDF wrapper, which is how you get files that seem impossibly large for their content.

The settings to change before you scan

If the document is still in front of you, fixing it at the source is far better than compressing afterwards, because you never create the bloat in the first place.

• Set resolution to 300 DPI for text documents. This is the standard for archival scanning and for optical character recognition, and going higher gains nothing you can see.
• Choose greyscale for anything that is not genuinely colour. Black text on white paper does not need colour channels.
• Avoid the aggressive black-and-white or 'text' modes on many scanners. They produce tiny files and tend to break thin strokes, which damages legibility and hurts OCR accuracy.
• If your scanner offers a PDF compression or quality setting, choose medium rather than maximum quality. Maximum is designed for photographs.

Fixing a file you already have

When rescanning is not an option — the document has gone back, or somebody else scanned it — compression is straightforward and effective. Scanned documents routinely shrink by seventy to ninety percent while staying entirely readable, because you are removing resolution that was never useful.

Start at a moderate setting rather than the most aggressive one, and check the result before sending it. The usual mistake is compressing to the smallest possible file when a middling setting would have cleared the limit comfortably and looked much better.

If you are combining several scans, merge them first and compress once at the end. Compressing each file separately and then merging pays the quality cost repeatedly for a worse final result.

The trade-off worth understanding

Compressing a scan re-renders each page as an image at a lower resolution. Since the page was already an image, you are not losing anything structural — but you are permanently discarding detail, and it cannot be recovered.

So keep the original scan. The compressed version is a copy made for one recipient with one size limit; the next person may need to read the small print, and regenerating a high-quality version from a heavily compressed file is not possible.

If you need the document to be searchable as well as smaller, run optical character recognition before compressing rather than after. OCR works from the image, and it works better on the higher-quality version.

Making the document searchable while you are at it

A scanned PDF has one other limitation people notice later: you cannot search it. There is no text in the file, so pressing find returns nothing no matter what you type, and a folder of scanned correspondence is effectively a filing cabinet you have to go through by hand.

Optical character recognition fixes this by examining the image, recognising the shapes as characters, and adding an invisible text layer beneath the picture. The document looks identical and becomes searchable, selectable and readable by screen readers.

Do it before compressing rather than after. Recognition accuracy depends heavily on image quality, so running it on the full-resolution scan gives a noticeably better result than running it on a version you have already squeezed. The text layer itself adds almost nothing to the file size.

A quick reference for common limits

It helps to know roughly what you are aiming at before you start, because the right setting depends on the destination.

As a rule of thumb, a single page scanned at 300 DPI in greyscale and compressed sensibly lands somewhere around 100 to 300 KB. If your pages are coming out at several megabytes each, the scanner settings are the thing to change, not the compression afterwards.

• Email attachments: most servers stop somewhere around 25 MB, and many corporate ones lower.
• Government and examination portals: frequently 1 or 2 MB, sometimes a few hundred kilobytes.
• Job application systems: commonly 2 to 5 MB for a CV with supporting documents.
• Court and legal filing systems: vary widely, and often specify a per-document rather than per-submission limit.

Tools mentioned in this guide