Compress PDF

A PDF is a bundle of content streams — text, fonts, images, vector graphics, form fields, annotations, metadata — packaged as objects inside a file container. Compressing a PDF means reducing the size of one or more of these objects, either by re-encoding them with a more efficient algorithm or by removing objects that are unused or duplicated.

The single largest driver of PDF file size in most documents is embedded images. A 100-page report with 50 photos at 300 DPI can easily reach 25-50 MB, while the same report with text alone would be under 500 KB. Because images dominate, most PDF compression strategies focus first on re-encoding images at lower quality or resolution.

Font embedding is the second-largest driver. When a PDF embeds a full font (as many documents do to guarantee glyph availability), it can add 200-500 KB per font per document. Subsetting fonts — embedding only the glyphs actually used — typically reduces embedded-font size by 80-95%.

PDF compressors typically use three strategies in combination:

Lossy compression (image re-encoding, downsampling) achieves the biggest size reduction — a 50 MB photo-heavy PDF might shrink to 5-10 MB — but introduces visible artefacts at aggressive settings. JPEG quality below 60 produces noticeable blocking on photos; below 40 the compression is often more distracting than the original file was.

Lossless compression (font subsetting, object flate compression) preserves the original visual output pixel-perfectly. Reduction is more modest — usually 20-40% of the original size — but the compressed PDF is indistinguishable from the original for reading purposes.

The right choice depends on the use case. For email attachments and web sharing, aggressive lossy compression at quality 75-80 is usually fine and cuts file size dramatically. For print production, contract exhibits, and archival copies, lossless is the safer default. Always keep an untouched original — compression is not reversible.

A 50 MB PDF might compress to 5 MB (90% reduction) or to 40 MB (20% reduction) depending on what is inside. The largest reductions come from PDFs that were poorly authored: PDFs exported from scanned documents at higher DPI than needed, PDFs assembled from JPG pages without image compression, PDFs from Word documents with embedded high-resolution photos.

Well-authored PDFs — those from LaTeX, from print production workflows, or from properly-configured Adobe Distiller — are already efficient and often show only 15-30% reduction from further compression. There is a floor below which further compression starts degrading visible quality.

A useful rule: if your PDF file size in MB is more than 3× the page count, there is significant compression headroom. If file size is under 100 KB per page, the file is already well-compressed and additional passes will not help much.

PDF/A is a subset of the PDF specification designed for long-term archival. It restricts several features that could interfere with future readability: no JavaScript, no encryption, no external content references, and mandatory font embedding. PDF/A-1 (2005) is based on PDF 1.4; PDF/A-2 (2011) and PDF/A-3 (2012) add JPEG 2000 and attachments respectively.

PDF/A compression uses only lossless techniques by requirement. Because font embedding is mandatory, PDF/A files are typically larger than equivalent standard PDFs — but the trade-off is guaranteed forward compatibility with PDF readers 30-50 years from now, which matters for legal records, government archives, and academic institutional repositories.

Compressing to PDF/A specifications requires a compliant workflow. Most consumer compressors (including ours in aggressive mode) may violate PDF/A rules. Use PDF/A specifically when archival compliance is a requirement.

Common email attachment ceilings in 2026: Gmail 25 MB, Outlook 20 MB (up to 34 MB via OneDrive links), corporate Exchange typically 10-25 MB depending on IT policy. A 30 MB PDF blocked by Gmail is a common practical need for PDF compression.

The best-effort target for reliable email delivery is under 10 MB — leaves room for the base64 encoding overhead that inflates attachment size by roughly 33% during transmission, and safely under most spam-filter attachment ceilings. Files above 25 MB should typically be shared via a link (Google Drive, OneDrive, Dropbox) rather than as attachments.

For contract signing workflows (DocuSign, Adobe Sign, HelloSign), the signing platforms typically accept up to 25-50 MB but slow noticeably above 10 MB during upload. Signed documents also gain 50-500 KB of certificate metadata; compress before signing where possible.

Over-compressing a document intended for print. If the PDF will be printed at 300 DPI on paper, downsampling images to 96 DPI leaves visible pixelation on print. For print output, keep images at 300 DPI at print size (or use lossless compression only).

Compressing a PDF that has already been compressed. Applying aggressive compression twice to the same PDF re-encodes JPEGs on top of already-compressed JPEGs, doubling the artefact accumulation. Always compress the ORIGINAL, not a previously-compressed version.

Not testing on the target device. A compressed PDF that looks fine on a desktop screen may show visible artefacts on high-DPI phone screens or at print resolution. Preview at the target resolution before distributing.

Losing OCR text layers during compression. If your PDF was OCR'd (searchable-text layer added), some compression tools strip the text layer during aggressive re-encoding. Verify text search still works on the compressed output before deleting the original.