How to Compress a PDF Without Losing Quality

Key takeaways

• Text and vector content compress losslessly — resolution settings never affect them.
• Image downsampling to 150 DPI is usually invisible on screen; 72 DPI is only safe for web-only documents.
• Removing embedded fonts you don't need can shave megabytes off a presentation-style PDF.
• Scanned PDFs are images all the way down — compression always trades file size against legibility.

A PDF that’s 40 MB when it should be 4 MB costs you every time: it’s slow to upload, rejected by email attachments, and annoying to share. But the wrong kind of compression costs you too — a form that compresses text into an unreadable blur, or a presentation where charts turn to mush.

The difference between a good compression result and a bad one usually comes down to understanding what’s actually inside the PDF.

What’s inside a PDF

A typical PDF is not one thing — it’s a container for multiple types of content that compress differently:

• Text is stored as character codes and positioning data. It’s already efficient and compresses further with lossless methods. Resolution settings have no effect on text.
• Vector graphics (charts, diagrams, logos created in Illustrator or similar) are stored as mathematical curves. Same as text — resolution is irrelevant, and lossless compression works well.
• Raster images (photos, screenshots, scanned documents) are stored as pixel data. These are what “DPI” and “image quality” settings actually affect, and they’re where most file size lives.
• Embedded fonts are a copy of the typeface data. A document with many embedded font families can be significantly larger than the content alone.

Knowing which type of content dominates your PDF tells you where to focus.

What “DPI” actually controls (and what it doesn’t)

When a PDF tool talks about reducing images to 150 DPI or 72 DPI, it’s resampling the raster images in the document to fewer pixels. Fewer pixels means less data, which means smaller file size.

150 DPI is generally safe for on-screen viewing and printing at standard sizes. Most people cannot visually distinguish a 150 DPI photo from a 300 DPI one on a monitor. If the document is going to be read digitally — emailed, uploaded, filled out on screen — 150 DPI rarely causes noticeable quality loss.

72 DPI is screen resolution for older monitors. It can make photos look soft or pixelated when zoomed, and it’s visually obvious in print. Use it only for documents that will only ever be viewed at 100% on screen and never printed.

300 DPI is standard for print. If your PDF contains photos that need to survive a printer at A4 or letter size, don’t go below this.

One thing that trips people up: if a PDF contains only text and vectors, changing the DPI setting does nothing — there are no pixels to resample. This is why compressing a text-heavy PDF with “reduce images to 72 DPI” sometimes produces zero file size reduction.

Font subsetting vs. embedding

If a document embeds the full typeface data for every font it uses, the file can be large even with no images. A Word document converted to PDF, for example, often embeds complete font files.

Font subsetting tells the PDF to include only the characters from the font that actually appear in the document, rather than the complete typeface. For Latin-alphabet documents, this typically removes 80–95% of the font data. The text renders identically because all the characters you need are still present.

If you’re seeing large PDFs from documents that don’t contain many images, embedded fonts are often the culprit, and subsetting is the fix.

Scanned documents: a different problem

A scanned PDF is entirely different from a text PDF. Every page is a raster image — a photograph of paper. There is no text layer, no vectors, no embedded fonts to trim.

This means:

• Resolution settings have a direct, immediate effect on quality and legibility.
• Compressing a 300 DPI scan to 72 DPI on a document with fine print will make some text unreadable.
• The right setting depends on the content: 200 DPI is often the minimum for readable body text; 300 DPI for anything with small print or handwriting.

For scanned documents, adding OCR (optical character recognition) during compression converts recognized text to real text, which compresses far better than pixel data. A 20 MB scan can sometimes become a 2 MB searchable PDF this way, with no visible quality loss in the text, because the text is no longer stored as pixels.

A practical compression order

For most documents, try this sequence:

1. Enable font subsetting — almost always safe, often the biggest gain for text-heavy files.
2. Resample images to 150 DPI — safe for digital use, significant size reduction for photo-heavy documents.
3. Run lossless recompression on existing image data — removes redundancy without changing pixels.
4. Remove metadata and annotations you don’t need — document history, author fields, comments, and print previews all add bytes.

If the result is still larger than you need, step down image quality — but check the output at 100% zoom before sending.

What to watch out for

Some tools compress aggressively by default in ways that aren’t obvious. Watch for:

• Forms with flattened fields — some compressors remove interactive form elements and replace them with pixel images of the filled-out form. This makes the file slightly smaller but the form no longer fillable.
• Transparency flattening — PDFs with transparent overlays (common from design tools) sometimes get flattened incorrectly, affecting how graphics appear.
• Color space conversion — compressors that convert RGB to CMYK or vice versa can shift colors noticeably, especially in branded documents.

If accuracy matters, compress in a tool that shows a preview before committing, and compare key pages at 100% zoom.