How to compress a PDF without ruining it

“Compress a PDF without losing quality” is one of the most searched PDF phrases, and the honest answer is uncomfortable: you usually cannot have both. Compression has to remove something — bytes, redundancy, or detail — and which one it removes decides what you lose.

The useful skill is not finding a magic setting. It is knowing which kind of compression you are getting, and choosing a target that matches what the file is for.

Two very different things called “compression”

Lossless compression

Lossless methods remove redundancy without discarding content: duplicate fonts get referenced once instead of embedded twice, unused objects are dropped, streams are packed more efficiently.

• Text stays selectable and searchable.
• Images keep their exact pixels.
• Typical savings: 5–20%, occasionally more on badly produced files.

If a file was exported from a word processor and contains duplicate font subsets, lossless compression can be genuinely free. On a scanned document, it barely moves the number.

Lossy compression

Lossy methods reduce image resolution and quality, or rasterise whole pages. This is where the dramatic savings come from — a 40 MB scan becoming 4 MB.

The cost is concrete:

• text is no longer selectable or searchable, because it is now a picture of text;
• small type and thin lines lose crispness;
• photographs show blocking and banding at aggressive settings.

Any tool promising “90% smaller with no quality loss” is doing lossy work and describing it generously.

Picking a target that matches the purpose

Work backwards from where the file is going, not from a percentage.

Purpose	Sensible target	What to accept
Email attachment	under ~20 MB	Some softness is fine; readability is not
Web download / upload portal	1–5 MB	Screen-quality images; text should ideally stay text
Internal archive	as small as possible	Loss is acceptable if the original is kept
Print, contracts, drawings	do not rasterise	Keep vector text; use lossless only
Legal or compliance record	original untouched	Store the compressed copy separately

If the document must remain searchable — contracts, filings, anything you will need to find text in later — avoid rasterising it. A 6 MB file with selectable text is more useful than a 1 MB file you cannot search.

Step by step

1. Check what kind of PDF you have. Open it and try to select a sentence. If text highlights, the document has real text and rasterising will destroy that. If nothing selects, it is already images — compression will only touch resolution.
2. Note the current size and the size you actually need.
3. Pick a quality setting rather than aiming for a percentage. Most tools expose a low / medium / high choice; medium is the usual compromise.
4. Download and compare side by side. Zoom to 100% on a page with small text. Check a page with a photo or a signature.
5. Keep the original until you have confirmed the compressed file is acceptable.

That last point is not bureaucracy. Compression is one-way in practice: you cannot recover detail from a rasterised page.

What compression does not fix

A file that is large because of embedded fonts. Subsetting fonts is a lossless repair; re-encoding pages is not the right tool.

A file that is large because of one enormous image. Compressing everything to fix one page is wasteful. Extract that page, re-encode only it, and merge the document back.

A file that is large because it contains hidden layers or attachments. Compression will not remove them; you need to inspect and strip them.

A file that is encrypted. You cannot re-encode an encrypted document without first removing the protection.

A quick way to tell text from pictures

Open the document and press Ctrl+F (Cmd+F on a Mac). Search for a word you can see on the page. If the reader finds it, the text is real text and compressing by rasterising will throw that away. If the search fails, the page is a picture — which is common for anything that came out of a scanner or a phone camera.

This one test tells you which kind of compression is even available to you, and it takes ten seconds.

The private way to do it

Server-based compressors upload the whole document, re-encode it elsewhere, and send it back. For a client contract or a medical record that is a real transfer of custody, however brief.

The compress tool here runs the re-encoding inside your browser. Two practical notes before you use it:

• Because the work happens locally, very large files are limited by your device memory rather than by bandwidth.
• The output is lossy by design. Read the result before deleting the original.

If you want to confirm nothing is transmitted, open the Network tab and run a compression — no request will carry your document.

Where to go next

If the reason you are compressing is an upload limit on some other service, what happens when you upload a PDF is worth five minutes first. If you need to work with no connection at all, see using PDF tools offline.