PDF compression: how it actually works

A PDF that ballooned to 40MB is almost never 40MB of text. Text compresses extremely well and rarely accounts for much file size on its own. The bulk is nearly always images — scanned pages saved at print resolution, screenshots pasted in at full size, or photos embedded without any downsampling.

What compression actually does

• Image downsampling — reducing the pixel dimensions of embedded images to match what the page will actually display, since a photo saved at 4000px wide is wasted on a page that displays it at 600px.
• Image re-encoding — recompressing images with a more aggressive JPEG quality setting, trading some visual fidelity for file size.
• Font subsetting — keeping only the glyphs actually used in the document instead of embedding the entire font file.
• Removing redundant data — duplicate objects, unused embedded resources, and metadata bloat that accumulates from repeated edits in different PDF editors.

The trade-off is almost always about images. Text and vector graphics (like charts made in the PDF itself, not pasted-in images) can be compressed losslessly, meaning no visible quality change. Photos and scans are where you have to pick a target: how much visual degradation is acceptable for how much size reduction.

When aggressive compression is fine

If the PDF is being emailed, uploaded to a portal with a size limit, or read on a screen rather than printed, moderate-to-aggressive compression is usually invisible to the reader. Scanned text documents in particular compress very well without looking noticeably worse, since the eye is reading text shapes, not fine photographic detail.

When to be more careful

Documents that will be printed, especially ones with photos or detailed diagrams, are more sensitive to over-compression — artifacts that are invisible on a phone screen become obvious on paper. If a PDF is being used as a legal or archival record, it's worth checking your organization's requirements before compressing at all, since some workflows require the untouched original to be preserved.

Processa's PDF compressor runs the downsampling and re-encoding steps and gives you a size estimate before you commit, so you're not guessing.