How to Compress a PDF Without Losing Quality

2026-08-11 · 8 min read

Most PDF compressors quietly turn your text into pictures. Here is what actually makes a PDF large, the two very different kinds of compression, and how to pick the right one.

You need to email a PDF and the attachment limit is 10MB. Your file is 34MB. You run it through a compressor, it comes back at 4MB, and you send it — without noticing that the text is no longer selectable, search returns nothing, and a screen reader would find an empty document.

That trade-off is invisible in most tools. Understanding it is the difference between compressing a PDF properly and quietly ruining it.

What is actually taking up the space

A PDF is a container holding several distinct kinds of data, and they contribute wildly unevenly to file size.

• Embedded images — usually 80 to 95 percent of a large PDF. A single uncompressed scan can exceed the entire rest of the document.
• Embedded fonts — 50KB to several megabytes each. A document embedding six full font families carries real weight.
• Text and vector content — almost always negligible. Thousands of pages of pure text is a small file.
• Metadata, thumbnails, revision history — small individually, but they accumulate in files edited many times.

This distribution matters because it tells you where compression can actually help. If your PDF is 30MB of scanned pages, no amount of metadata cleanup will meaningfully shrink it. If it is a 5MB text report, the images are not the problem.

Two kinds of compression that get called the same thing

The word "compress" covers two fundamentally different operations, and tools rarely distinguish them.

Structural compression rewrites the file's internal plumbing. It strips metadata, removes orphaned objects left behind by editing, deduplicates repeated resources, and packs the remaining objects into compressed streams. Nothing visible changes. Text stays selectable and searchable. Typical saving is 5 to 20 percent — occasionally much more on files that have been edited repeatedly.

Image re-encoding decodes every embedded image and re-encodes it at lower quality or resolution. This is where the large savings live — 40 to 80 percent is routine. But depending on implementation, it may rasterise entire pages, converting your text into a picture of text. That is what breaks selection, search and accessibility.

Which one you need

Use structural compression when the document is text-heavy and the text matters: contracts, reports, academic papers, anything that will be searched, quoted or read by assistive technology. Accept the smaller saving.

Use image re-encoding when the document is image-heavy and visual fidelity at screen resolution is all you need: scanned documents, brochures, product catalogues, photo portfolios. The text in a scan is already an image, so you lose nothing that was not already lost.

Our PDF compressor makes you choose between these two modes explicitly rather than deciding for you, and tells you what each one costs.

The honest limits of browser-based compression

Desktop tools built on Ghostscript, and commercial products like Acrobat, can do things a browser cannot. They subset embedded fonts to include only the glyphs actually used, which can reclaim megabytes on font-heavy documents. They recompress images with format-aware heuristics, choosing JPEG for photographs and lossless formats for line art within the same file.

Browser-based tools work with what the JavaScript PDF libraries expose. Structural compression and image re-encoding are both well supported. Font subsetting is not. If you have a 50MB PDF that is mostly embedded typefaces, a desktop tool will beat any browser tool, and you should use one.

What the browser gives you in exchange is that the document never leaves your machine — which for a contract or a medical record is often the deciding factor.

A practical sequence

• Check what you are dealing with — try selecting text in the PDF. If you cannot, it is already a scan and image compression costs you nothing.
• Try structural compression first. If the file was heavily edited, this alone sometimes solves the problem.
• If you need more, switch to image compression and start at the mildest setting.
• Open the result and read it at the size people will actually view it. Compression artefacts that are glaring at 400 percent zoom are invisible at 100 percent.
• Confirm text is still selectable if that mattered to you.

When compression is the wrong answer

If your PDF is large because it contains 200 pages and you only need to send 12, extracting those pages shrinks the file far more than any compressor and costs you no quality at all.

If you are sending several documents, splitting them and sending separately may clear an attachment limit without touching quality. And if the recipient only needs to read the content rather than keep the file, a link is better than an attachment.

← Back to all articles