How to compress a PDF without losing quality

"Compress this PDF but keep it readable" is one of those requests that sounds simple and turns out to hide a dozen decisions. A PDF can shrink a lot or a little depending on what's inside it, and the difference between a clean 2 MB file and a blurry 2 MB file comes down to which parts you compress and how hard. This post walks through what actually makes a PDF big, which knobs matter, and how to get the size down without turning your document into mush.

First, find out why the file is big

PDFs are containers. A single file can hold text, vector graphics, embedded fonts, raster images, and metadata, and any one of those can dominate the size. Before you compress anything, it helps to know what you're dealing with:

• Scanned documents are almost always images. Every page is a photograph of paper, so the file is heavy and the text isn't even selectable.
• Exported design files (from InDesign, Illustrator, or a slide deck) often carry high-resolution images meant for print, at 300 DPI or more, when the screen only needs 96 to 150.
• Text-only reports are usually small already. If one is large, the cause is often embedded fonts or a few stray high-resolution logos.

The category tells you where the savings are. You don't compress a text report the same way you compress a 40 MB scan.

The knob that matters most: image downsampling

In most heavy PDFs, images are the cost. Compression tools reduce image weight in two ways, and it's worth knowing the difference because they affect quality differently.

Downsampling lowers the resolution. A photo stored at 300 DPI gets resampled to, say, 150 DPI. Fewer pixels means a smaller file. As long as the target DPI matches how the document will be used (on screen, 150 is plenty), you lose nothing a human eye will notice. Downsample below what the display needs and you start to see soft edges.

Re-encoding changes how each image is stored. JPEG compression has a quality setting from 0 to 100. Drop it from 95 to 75 and the file shrinks noticeably while staying clean. Drop it to 40 and you get visible blocky artifacts around text and sharp edges. The sweet spot for most documents is somewhere between 70 and 85.

Why "keep it readable" usually means "leave the text alone"

Real text in a PDF (the kind you can select and copy) costs almost nothing. It's stored as characters plus font information, not as pixels. Compressing text gives you tiny savings and risks nothing, so good compressors leave it untouched. The mistake people make is running a scanned PDF, where the "text" is actually an image, through an aggressive image setting. That blurs the very thing they wanted to keep sharp.

If your document is a scan and you need both small size and crisp text, the real fix is OCR plus re-rendering, which is a heavier operation than plain compression. For a normal PDF with selectable text, you get the best of both: shrink the images, keep the text perfect.

Fonts: the silent weight

PDFs embed fonts so the document looks the same everywhere. That's good for fidelity and bad for size when a file embeds full font families it barely uses. Font subsetting fixes this. It strips out every glyph the document doesn't actually contain, so a file using twelve characters of a decorative font stops carrying the other few thousand. Most modern export tools subset by default, but older files and certain workflows skip it, and the savings can be surprising.

A practical order of operations

1. Identify the file type: scan, design export, or plain text.
2. If images dominate, downsample to the resolution you actually need (150 DPI for screen, 200 to 300 for print).
3. Set JPEG quality in the 70 to 85 range. Check one image-heavy page before committing.
4. Subset fonts if the tool offers it.
5. Leave selectable text and vector graphics alone.

Doing it without uploading the file

Most online compressors send your document to a server, run a tool like Ghostscript, and send it back. That works, but the file leaves your machine. For anything private (contracts, medical records, internal reports) that's a real cost. PDFShore runs compression in the browser instead. It reads the file locally, downsamples and re-encodes the images on your device, and writes a new PDF without ever uploading the original. You get the size control described above with none of the server round trip.

The honest trade-off: a browser uses your machine's memory, so a very large file on an old laptop will be slower than a beefy server would be. For the vast majority of documents, the difference is a second or two, and the file never leaves your control.

What good compression looks like

A well-compressed PDF is one where you can't tell it was compressed. Text stays sharp because you never touched it. Images look fine at the size they're displayed because you matched resolution to use. The file is a fraction of the original because the waste (overscale images, unused glyphs, redundant data) is gone. That's the goal: smaller for the right reasons, not smaller at the cost of looking cheap.