How to reduce PDF file size without losing quality

Most advice about shrinking a PDF skips the first question you need to answer: what is actually taking up the space? Two files with the same number of pages can differ in size by a factor of fifty, and the method that fixes one will do nothing for the other. Spend five minutes finding out which kind you have, and you will save an hour of trial and error.

First, work out what kind of PDF you have

Open the file and try to select a sentence with your cursor. What happens next tells you almost everything.

• The text highlights normally. This is a text-based PDF. The words are stored as characters plus font data. These files are already small (usually well under 100KB per page), and there is very little left to squeeze out.
• Nothing highlights, or a whole page selects as one block. This is a scan: every page is a photograph. This is where the weight is, and where compression makes a big difference.

A 40MB file is almost never 40MB of text. It is images: scans, screenshots, product photos, or a logo someone dropped in at 4000 pixels wide and shrank to an inch.

What actually makes a PDF large

In rough order of how much space each one wastes:

1. Scanned pages at unnecessary resolution. Scanners default to 600dpi. For text you will read on a screen, 150dpi is fine; for text you will print, 300dpi is plenty. Going from 600 to 200dpi cuts the pixel count by nine.
2. Colour scans of black-and-white documents. A greyscale scan is roughly a third the size of a colour one, and a page of black text on white paper has no colour in it to lose.
3. Embedded images at their original size. A 12-megapixel phone photo placed into a document keeps all 12 megapixels even when displayed two inches wide.
4. Fully embedded fonts. Each embedded font adds a few hundred kilobytes. That is fine for a document that must look identical everywhere, and wasteful for an internal memo.
5. Revision history. PDFs can be saved incrementally (the ISO 32000-1 specification calls these incremental updates), keeping every previous version inside the file. Re-saving the document flat discards all of it.

Choosing a compression level

Compressing a scanned page means re-encoding the page image as a smaller JPEG. That is a real trade: fewer bytes, less detail. The right level depends on what happens to the document next.

Use	Target	Expect
Email attachment, read on screen	Strong, greyscale	80–95% smaller
Shared internally, occasionally printed	Balanced	50–80% smaller
Client-facing, or legal filing	Light	20–50% smaller
Archival master copy	Do not compress	No change

Always work from a copy, and always look at the output before you delete the original. Text that was crisp at 300dpi can turn mushy at 100. Your own eyes, on your own document, have to decide whether that is acceptable.

Compressing without losing the text

The usual complaint about compressing a PDF is that it stops being searchable. That happens when a tool rasterises every page and throws the text layer away. You get a smaller file that is now a stack of pictures, and Find no longer works.

Our Compress PDF tool re-encodes the page images but keeps the original text as an invisible layer, so the document stays searchable and selectable at the smaller size. If the file was pure text to begin with, it will tell you that compression would make it larger rather than quietly handing you a bigger file.

When compression is the wrong tool

• The file is large because it has 400 pages. Compressing will barely help. Split it and send the part that is actually needed.
• You need the images, not the layout. Convert straight to JPG at the resolution you want.
• You need the text, not the pages. Extract it to plain text or Word and the file becomes a rounding error.
• The problem is one enormous image. Compress that image on its own and rebuild the document.

Four ways to shrink a PDF, and when each one wins

Each of these is a different operation. When a file refuses to get smaller, the usual reason is that the wrong one was picked.

Method	What it changes	Keeps text selectable	Use it when	Fails when
Compress to a target size	Tries settings until the file fits the number you typed	Yes	A portal or form gave you an exact limit	The document is too long for that limit at any readable quality
Compress to a quality level	One fixed setting: strong, balanced or light	Yes	You want it smaller and have no exact ceiling	You need to land under a specific number, because presets overshoot or undershoot
Split into parts	Nothing about the pages; there are simply fewer per file	Yes	A long scan cannot reach the limit legibly	The recipient needs one document
Re-scan or re-export	Produces fewer pixels in the first place	Depends on the scanner	You still have the original document or source file	The PDF is all you have

One rule saves a lot of wasted attempts. If the PDF is text with no scanned pages, neither of the first two will help much, because there are no images to shrink. Compressing it can even make it slightly larger. Split it or re-export it instead.

Common questions, answered in one line each

• Why did my PDF get bigger after compressing? It was already text-only, so the compressor re-encoded structure that was optimal to begin with and added its own overhead.
• Can I compress a PDF without losing quality at all? Only by removing what is not visible, such as duplicate fonts, revision history and unused objects. Real reductions on a scan always cost some detail.
• Why will my 12-page scan not reach 100 KB? That is about 8 KB a page. Text stops being readable somewhere around 25 KB a page, so the real fix is to split the file or scan at a lower setting.
• Does compressing a PDF remove its password? No. An encrypted PDF has to be unlocked before anything can read the pages at all.
• Will compression break a signed or filled form? It can. Flattening and re-encoding a page can drop form fields and invalidate a digital signature. Keep the original.

A practical sequence

1. Try to select text. Scan or not? That decides everything else.
2. If it is a scan, compress at Balanced and check the result.
3. Still too big? Try greyscale. On a text document you will lose nothing you care about.
4. Still too big? The file is long, not heavy. Split it.
5. If it is text-based and still large, re-save it flat to drop the revision history.

Every one of those steps runs in your browser on this site, which means you can do all five to a confidential document without it being uploaded anywhere at any point.