How to reduce a PDF's file size (and why some files won't shrink)
What actually makes PDFs big, how the three compression levels work, and how to choose the right one for email, web or print.
A 40 MB PDF that bounces from an email server is almost never 40 MB of text. It is 40 MB of images: photographs embedded at full camera resolution, scanned pages stored as uncompressed bitmaps, or a logo pasted into every page header at 300 dpi. Understanding this makes it obvious what compression can and cannot do.
What takes up space in a PDF
- Images — by far the largest contributor. A 300 dpi colour scan of an A4 page is roughly 25 megapixels before compression.
- Embedded fonts — usually a few hundred kilobytes per font, and every subset counts. Documents assembled from many sources can carry dozens.
- Unused objects — old page versions, orphaned images and form fields left behind by repeated editing. Word processors are notorious for this.
- Uncompressed streams — some generators write content streams and metadata without any compression at all.
Text and vector drawings are tiny by comparison. A 200-page text-only report is often under 2 MB and cannot be compressed much further.
How the compression levels work
PDFCraft's compressor runs on the server, because it uses Ghostscript, a full PDF interpreter that re-encodes every image. Three levels are available:
| Level | Colour/grey images | Black-and-white scans | JPEG quality | Best for |
|---|---|---|---|---|
| Low | 150 dpi | 300 dpi | 85 | Printing, archives |
| Balanced | 110 dpi | 300 dpi | 75 | Email attachments, sharing |
| High | 72 dpi | 200 dpi | 60 | Web viewing, previews |
Every level also removes unreferenced objects and rewrites the file structure with object streams. In parallel, a lossless pass is tried with qpdf, which recompresses streams without touching image quality. The smallest valid result wins.
Why some files don't get smaller
If your file is already text-only, or was produced by a tool that compressed images aggressively, none of the candidates will beat the original. In that case PDFCraft returns your original file unchanged and tells you so. We think that is better than the alternative some tools choose — silently handing back a "compressed" file that is larger or visibly degraded.
Tips for the best result
- Start with balanced. It cuts most scanned documents by 60–85% with no visible difference on screen.
- Use low for anything that will be printed. 150 dpi is the threshold below which photographs start to look soft on paper.
- Check the saving before you download. The result card shows the original size, the new size and the percentage saved.
- Compress before merging, not after, when you are combining many large scans: it keeps each browser-side merge fast.
- Scans still huge after compression? They were probably saved as lossless PNG or TIFF inside the PDF. High compression converts them to JPEG, which is where the real saving is.
Files sent to the server are deleted automatically after the retention period shown on the tool page — normally two hours — and you can delete them earlier from your dashboard.