PDF & documents

Why Your PDF Is Too Large for the Upload Limit — and How to Fix It

What actually makes a PDF big, how to diagnose the cause, and which compression approach to use for scans, text documents and image-heavy files without destroying legibility.

8 min read · Updated 2026-07-08

A three-page PDF should not be 28 MB. When it is, the PDF format is not to blame — something inside it is, and until you know what, compression is guesswork. This guide explains the four things that make PDFs large, how to tell which one you are dealing with, and what to do about each.

A PDF is a container, not a compression format

This is the misconception behind most confusion. A PDF does not compress your content — it packages it. Inside are separate objects: fonts, vector drawing instructions, embedded images, metadata, and sometimes entire attached files. Each is stored with its own encoding. The PDF's size is essentially the sum of its contents.

So "compressing a PDF" really means recompressing the things inside it, and what works depends entirely on what those things are.

The four causes

1. Embedded images at full camera resolution

By far the most common cause. If you built the PDF from phone photos, each page carries a 12-megapixel image of roughly 3–5 MB. Ten pages is 40 MB. The page displays at A4 size where about 2 megapixels is more than sufficient, so roughly 80% of that data is doing nothing.

Diagnosis: the PDF came from a scanner app, a phone camera, or Image to PDF. You cannot select any text in it.

Fix: downsample the images. Either resize the source images to about 1800 pixels on the long edge and rebuild the PDF, or run the existing file through the PDF Compressor.

2. Images stored losslessly

Some tools embed page images as PNG or uncompressed bitmaps. A photograph stored as PNG can be ten times larger than the same image as JPEG with no visible benefit.

Diagnosis: the file is enormous relative to its page count, and zooming in shows a clean photographic image with no JPEG artifacts at all.

Fix: re-encode as JPEG. Our compressor does this as part of its normal operation.

3. Fully embedded font families

PDFs embed fonts so the document renders identically everywhere. Normally only the glyphs actually used are included — a technique called subsetting. Some exporters skip subsetting and embed the complete font, and a full CJK or Devanagari font family can run to several megabytes on its own.

Diagnosis: a text-only document with no images that is nonetheless several megabytes.

Fix: re-export from the source application with font subsetting enabled. This one is best solved upstream — image-based compression will not help and will make your text non-selectable for nothing.

4. Accumulated revision history

PDF supports incremental saving: each edit appends to the file rather than rewriting it. A document edited many times can carry every previous version inside it, including content that was supposedly deleted.

Diagnosis: a document that has been through many rounds of editing and is far larger than its visible content justifies.

Fix: any operation that rewrites the file completely will discard the history. Running it through a compressor or a merge/split cycle does this.

Choosing a compression strategy

When compression is not the answer

Sometimes the file is large because it contains more than it needs to.

• Extra pages you do not need to submit. Use Extract PDF Pages to pull out only the relevant ones, or Delete PDF Pages to remove the rest. A 60-page prospectus where the portal wants page 12 is not a compression problem.
• Blank pages from a duplex scan of single-sided documents. Delete them — they cost real bytes.
• Multiple documents merged unnecessarily. If the portal has separate upload slots, Split PDF and submit them individually.
• A cover page or scanning-app watermark that adds nothing.

Removing pages is lossless with respect to the pages you keep. Always try it before compressing.

How small can you realistically get?

If a portal demands 500 KB for a 20-page scanned document, that is 25 KB per page and something will have to give. Consider whether all 20 pages are actually required, or whether greyscale is acceptable — dropping colour on a text scan typically saves 40–60%.

Always check the result

Open the compressed file and read it at 100% zoom before you submit. Specifically check the smallest text on the page, any handwritten annotation, stamp impressions and seals, and signatures. These are the first things to become illegible and the most likely to matter to whoever reviews the document.

Frequently asked questions

Why is my one-page PDF 15 MB?

Almost certainly a full-resolution photograph embedded at scanner-app defaults, possibly stored losslessly. Resize the source image to about 1800 pixels on the long edge and rebuild, or run it through the PDF compressor.

Does compressing a PDF reduce its page count or dimensions?

No. Page count and page size are unchanged. What changes is the resolution and encoding of the images inside each page.

Why can I no longer select text after compressing?

Because image-based compression re-renders each page as a picture. The characters are still visible but are no longer text objects. For scans this changes nothing, since there was no text layer to begin with. For a document exported from a word processor, re-export from the original instead.

Is it safe to compress an official document?

Technically yes — the content is unchanged, only the image encoding differs. Verify legibility afterwards, particularly of stamps and signatures. Some authorities require an uncompressed or original-quality scan; check before compressing anything that must be certified.

How do I password-protect a PDF here?

We do not offer that. Reliable client-side PDF encryption is not achievable with the libraries this site is built on, and shipping something that looks like protection but is not would be worse than shipping nothing. Every other PDF operation on the site is fully functional.

Are my PDFs uploaded to your servers?

No. All PDF processing runs in your browser using pdf-lib and pdf.js. The file is read locally, modified locally and downloaded locally. We have no server that could receive it.