How to actually compress a PDF (and why most tools can't)

A PDF isn't one thing

Think of a PDF as a ZIP archive with a table of contents. Inside: some text, some fonts, some images, a little bit of metadata, and a cross-reference index that points at the bytes. Each of those pieces is already compressed in its own way.

When a "compress PDF" tool runs, one of two things is happening:

1. Wrapper-compress — re-save the PDF with more efficient object streams, drop the xref index padding, strip title/author metadata. Savings: 5-20% on text-heavy PDFs with sparse images, 0% (or worse) on image-heavy PDFs.
2. Deep-compress — rasterize every page to pixels, re-encode the pixels as a lower-quality JPEG, rebuild the PDF around those JPEGs. Savings: 60-90% on image-heavy PDFs, at the cost of vector text becoming a bitmap.

Most free tools only do #1. When a user uploads a scanned document — which is literally one giant image per page — path 1 saves nothing. Honest tools tell you. Dishonest tools quietly hand back the same file with a "-compressed.pdf" suffix and hope you don't check.

Which path should you use?

Use wrapper-compress when…

• The PDF is mostly text (reports, ebooks, invoices)
• You need the text to stay selectable and searchable
• Print quality matters — rasterization introduces pixelation
• You want a fast operation (wrapper-compress is < 1 second)

Use deep-compress when…

• The PDF is a scanned document (every page is an image)
• The PDF embeds high-resolution photos or camera captures
• You care about file size more than text selection
• You're emailing the file and hit a 25 MB attachment limit

Our PDF compressor runs wrapper-compress by default (fast, safe, non-destructive). Tick "Re-encode images" in the settings panel to switch to deep-compress. Both paths run entirely on our servers — your file never hits disk.

Why your file might not have gotten smaller

The #1 failure mode on compression tools: the PDF is already well-compressed. Our tool detects this and, rather than handing you a useless output, it returns the original filewith a warning banner:

“This PDF is already well-compressed — we returned the original. Try the deep-compress option (re-encode images) for image-heavy PDFs.”

This happens when:

• The PDF was exported by modern software (MS Word 2013+, Pages, Google Docs) which already uses object streams
• The embedded images are already compressed (JPEG inside a PDF stays JPEG — we can't magically recompress it without re-encoding)
• Fonts are already subsetted (only the glyphs used are embedded)

What the "quality" slider actually does

On the deep-compress path, the quality slider (30-100) controls the JPEG quality of the rasterized page images. Lower = smaller file + more visible artefacts.

• 85 (default): visually identical to the source at normal zoom. Fine for most use cases.
• 70-80: slight softening visible at 100% zoom. Good for email attachments and web uploads.
• 50-60: visible compression artefacts on gradients and text edges. Use only when you have a hard size budget.
• Below 50: text legibility starts to suffer. Don't.

Size expectations by source

Rough savings on deep-compress at quality 85:

• Phone-camera scan of a document: 10 MB → 1-2 MB (80-90% reduction)
• PowerPoint export with photos: 8 MB → 2-3 MB (60-75%)
• Textbook chapter (scanned): 25 MB → 4-6 MB (75-85%)
• Text-only report (Word export): 2 MB → 1.6-1.9 MB (5-20% on wrapper-compress path)

If your savings are much smaller than this, the file was likely already compressed. Trying a third-party compressor won't help unless it rasterizes. They all run the same pdf-lib / iLovePDF-style wrapper path under the hood.

What about text-selection after deep-compress?

Deep-compress destroys text selection because the text becomes part of a bitmap. If you need searchable text AND small file size, the workflow is:

1. Deep-compress the PDF to shrink the images
2. Run OCR (optical character recognition) to re-add a text layer behind the images

We ship the first step. OCR is on our roadmap— it's one of the most-requested features. For now, pair our deep-compress with Adobe Acrobat's "Recognize Text" feature, or Tesseract if you're comfortable with a command line.

Common questions

Is it safe to upload my PDF?

Yes. Files are processed in memory and auto-deleted within the hour. No disk writes, no database, no analytics on file contents. See our privacy policy for the full details.

How big a file can I compress?

Free tier: 4 MB per file. Paid plans lift this to 1.5-5 GB per file depending on tier — see pricing. For files over 4 MB we use direct-to-storage upload so your browser doesn't have to hold the whole file in memory.

Can I batch multiple PDFs?

Yes — drop multiple PDFs on the compressor and we'll process them sequentially, then ZIP the outputs. Works on every plan.

Does this work for scanned receipts / tax docs?

Perfectly. Scanned documents are the scenario where deep-compress shines most. A 20 MB folder of receipts typically comes down to 3-5 MB at quality 85, still perfectly readable.

Ready?

Compress PDF →. Free tier works for files up to 4 MB. No sign-up.