Documents

How to optimise heavy PDF files for email without losing quality

Written and maintained by the KitYards team Published: 7 min read

The message comes back within seconds: attachment too large. The file is 31 MB, the limit is 25 MB, and the meeting is in ten minutes. The instinct is to run it through the first compressor available and hope the result is still legible. It usually is not, because indiscriminate compression attacks the wrong part of the document. Getting a PDF under an attachment limit reliably is a matter of understanding what is heavy in it, and treating that specifically.

The limits you are actually fighting

Every mail system enforces its own ceiling, and the smallest one in the chain wins. Gmail accepts attachments up to 25 MB. Outlook.com is also 25 MB, but a corporate Microsoft 365 mailbox is commonly configured lower — 20 MB is a frequent default, and administrators tighten it further. iCloud Mail sits around 20 MB, Yahoo and Proton around 25 MB. Many corporate gateways and antivirus scanners impose their own, stricter limit on top.

The critical detail is that these numbers do not apply to your file as it exists on disk. Email attachments are Base64-encoded so that binary data can travel through a text-based protocol, and Base64 inflates data by roughly 33%. A 24 MB PDF arrives at the server as about 32 MB of encoded payload, which is why an attachment comfortably under the published limit is still rejected.

Find out why the file is heavy before you compress it

PDFs get large for a small number of reasons, and each has a different remedy. Diagnosing first takes thirty seconds and saves you from destroying quality you did not need to sacrifice.

Scanned pages

This is the most common cause by a wide margin. A scanned document is not text at all: each page is a full-page photograph. Scanners frequently default to 300 or 600 DPI in colour, which produces 2 to 8 MB per page. A twenty-page scanned contract can trivially exceed 60 MB while containing less information than a one-page text document.

Print-resolution photographs in an office document

When a 12-megapixel phone photo is pasted into a report and shrunk to a quarter of the page, the layout shows it small but the file still stores every original pixel. Ten such images add 30 MB to a document whose text weighs a few hundred kilobytes.

Structural bloat

Embedded font subsets duplicated across generations of edits, incremental save history that retains superseded versions of pages, unused objects, form data, attachments and colour profiles all accumulate. On a document that has been passed around and re-saved a dozen times, this alone can account for a third of the size — and removing it costs nothing visually.

Lossless and lossy: what each one really does

Compression is not one operation. Two very different families of technique get the same label, and knowing which one you are applying is the difference between a clean 40% reduction and an unreadable document.

  • Lossless work rewrites the file's structure without touching the appearance: it recompresses content streams, removes unused and duplicated objects, discards superseded revisions, drops metadata and optimises the cross-reference table. Every pixel and every glyph survives exactly. Typical saving is 10–40%, and on a heavily edited document it can be far more.
  • Lossy work reduces the information in the images: downsampling them to a lower pixel density and re-encoding them as JPEG at a chosen quality. This is where the large reductions come from — 70–90% on a scan-heavy file — and also where quality is spent. Used carefully it is invisible; used bluntly it turns small text into a grey smear.

Compress a PDF in your browser

Rebuilds the document locally: structure and streams are optimised and images are re-encoded at the quality level you pick. Nothing is uploaded, so a confidential contract stays on your machine while you get it under the limit.

Compress a PDF

A workflow that gets under the limit without guessing

  1. 1 Establish the target. Find the recipient's limit if you can; otherwise assume 20 MB and divide by 1.4. Your target is roughly 14 MB.
  2. 2 Try lossless first. Run the document through a structural optimisation with images untouched. On a text-heavy or much-edited file this alone may finish the job, at zero cost to quality.
  3. 3 If it is still too big, downsample the images. For a document that will be read on a screen, 150 DPI is ample; 96–120 DPI is acceptable for reference copies. Only go to 300 DPI when the recipient will print it.
  4. 4 Re-encode photographs, not diagrams. JPEG quality around 75–80 is visually clean for photographic content. Line art, tables and screenshots suffer badly at those settings — if the document is mostly diagrams, downsample less and accept a larger file.
  5. 5 Check the result at 100% zoom. Not fitted to the window — at full size, on the pages with the smallest type. If footnotes and figures in tables are still crisp, you are done.
  6. 6 If it will not fit, stop compressing and split. Past a certain point you are destroying the document to satisfy a transport limit, which is the wrong trade.

When splitting is the better answer

A 90 MB scanned tender response will not become a 14 MB attachment while remaining useful. Sending it as three parts, each a coherent section, is more professional than sending one unreadable file — and often more convenient for the recipient, who may only need the appendix. Splitting is also the right move when only part of the document should be shared at all: extracting pages 12 to 18 discloses far less than sending the whole file and asking the reader to ignore the rest.

Split a PDF into sections or extract a page range

Choose the ranges you need and get them as separate files. Nothing is re-encoded, so each part keeps the original quality of the pages it contains.

Split a PDF

Resolution targets worth memorising

  • 96–120 DPI — reference copies and documents that will only ever be skimmed on a screen. Smallest files; small print starts to soften.
  • 150 DPI — the practical sweet spot for sharing. Sharp on any screen, including at zoom, and typically a third to a quarter of the size of a 300 DPI original.
  • 300 DPI — the standard for anything that will be printed on paper. Do not go below this if the recipient will produce a hard copy.
  • 600 DPI — archival scanning and documents destined for OCR of small or degraded type. Rarely justified for email.

Four things that make the file bigger

  • Printing to PDF to “flatten” it. This re-renders the document, frequently rasterising vector text into images. The result is larger, blurrier and no longer searchable.
  • Screenshotting pages and reassembling them. You convert compact vector text into full-page bitmaps, then compress those. It is the worst possible order of operations.
  • Zipping the PDF. A PDF's streams are already compressed. Zipping typically saves 1–3%, and some mail gateways block or quarantine archives anyway.
  • Round-tripping through images. Exporting every page to JPG and rebuilding a PDF from them destroys the text layer permanently and usually increases the size.

Quality control before you send

  • Open the compressed file, not the original, and read the two densest pages at 100%.
  • Confirm the text is still selectable — if you can no longer highlight a word, something rasterised the page.
  • Check that signatures, stamps and handwritten annotations are still legible; they degrade earlier than printed text.
  • Verify the page count matches, and that no pages were rotated or dropped.
  • Check the final size on disk against your target, not against the published limit.

Compress the images before they go into the document

If you are building the PDF yourself, resizing and compressing the photographs first is far more effective than compressing the finished file. A 4000-pixel photo shown at 800 pixels wide is 96% wasted data.

Compress images

When the attachment is simply the wrong channel

Some documents should not be emailed at all. Anything above roughly 20 MB, anything the recipient will need in its original quality, and anything you may need to revise are all better served by a shared link. Email attachments also have a habit of multiplying: five recipients means five copies stored indefinitely in five mailboxes, plus whatever the mail server keeps. If the content is sensitive, that alone is a reason to send a link with an expiry instead — and to compress the file for reading, not for transport.

The tools used in this guide

Free, no sign-up, and everything runs on your own device.

Keep reading