How to Compress a PDF Without Uploading It (Browser-Only, Free)
If your PDF is too big to email, the search results for "compress PDF" are a minefield. The top 10 sites — Smallpdf, iLovePDF, PDF24, Adobe's online tool — all do the same thing: you upload your file to their server, their server compresses it, you download the result. That's fine for a flyer. It's a problem for a contract, a tax return, a medical record, or anything you'd rather not have sitting on a stranger's S3 bucket for the next 30 days.
The good news: PDF compression is mostly recompressing the embedded images. That can happen entirely in your browser, with no upload at all.
Why PDFs get big in the first place
A PDF that's 30 MB is almost never 30 MB of text. It's typically:
- Embedded raster images at full resolution (a phone-camera shot is ~5 MB; embed 6 of them, you've got 30 MB)
- Scans where each page is a 300 DPI image instead of selectable text
- Embedded fonts (one full font face is ~200 KB; PDFs sometimes embed 4-5)
- Form data with embedded image stamps for signatures
Compression mostly attacks the first two — recompresses the embedded images at lower quality, downsamples them if they're absurdly oversized for the page they sit on. Text and fonts are basically untouched (and they're tiny anyway).
How to do it in your browser
Buncha's PDF compressor runs entirely client-side via pdf-lib. Drop the PDF, pick a quality preset (low / medium / high), download the result. The file never leaves your browser. Pull up the network tab — you'll see no requests during the compression.
Three quality levels:
- Low — heavy recompression, ~70-90% size reduction. Suitable for screenshots, drafts, or anything you'll print.
- Medium — moderate recompression, ~40-60% reduction. The default. Good for emailing contracts or invoices where the visual fidelity needs to be readable but not pristine.
- High — light touch, ~10-30% reduction. Use this when the PDF still has to look professional or be printed at quality.
The numbers depend heavily on what's in the PDF. A scanned document compresses dramatically (the embedded scans are huge); a text-only PDF won't budge much because there's not much to compress.
What can go wrong
Some PDFs resist client-side compression:
- Encrypted / password-protected PDFs can't be re-saved without the password. Use /tools/pdfunlock first if you have the password and need to remove it.
- Very large files (300+ MB) may exhaust browser memory. The desktop has more headroom than mobile here.
- PDFs with embedded JavaScript or unusual structures sometimes break on re-save. Modern pdf-lib handles 95%+ of consumer PDFs cleanly, but the long tail exists.
If a PDF won't compress, it's almost always because it's already as small as it can be — text-only PDFs from Word documents are typically 100-500 KB, and there's nothing to remove.
The privacy difference, concretely
Smallpdf's terms of service explicitly let them retain uploaded files for 60 minutes (free tier) and longer for paid. iLovePDF's privacy policy says files are deleted "after 2 hours." Adobe's free online tool requires an Adobe account and stores the file in their cloud.
Buncha's PDF compressor never sees your file. There's no privacy policy section about retention because there's nothing retained — the compression happens in JavaScript that runs on your machine. The same is true for merging PDFs, splitting them, signing, and redacting.
This isn't a feature we add to the marketing copy. It's a structural property of how the tools are built.