PDF Tools·4 min read

PDF Compression Explained: What Shrinks and What Doesn't

It's 11 pm, the portal closes at midnight, and your upload keeps bouncing because the PDF is 25 MB against a 10 MB cap. Before you start deleting pages, it helps to know what a compressor can genuinely remove, what it can't, and why two files with the same page count shrink so differently. This guide walks through the levels in the Compress PDF tool, the target-size mode, the Scanned document mode for stubborn scans, and the order in which to try them so you lose as little quality as possible.

Open the free Compress PDF All insights

The upload limit that starts most compression jobs

Upload limits don't negotiate.

Almost nobody compresses a PDF for fun. The trigger is a hard ceiling: a job portal that stops at 10 MB, an application form that wants 200 KB, or an email server that bounces anything over 25 MB. Your document is finished, correct, and slightly too heavy, and the deadline doesn't care.

The right move depends on what's inside the file. A 25 MB PDF built from scanned pages and a 25 MB PDF exported from Word are different animals, and the same setting treats them very differently. Knowing which one you're holding saves you an hour of re-uploading.

What each compression level does

Structure first, then pictures.

A PDF is a container of objects: fonts, pages, images, metadata and an index that ties them together. At every level, the tool re-saves the file with object streams, which packs loosely written objects together. The Lossless level stops there. Nothing visible changes, and the saving is usually small.

Light, Medium and Strong also re-encode the images inside the PDF as JPEG, at 80, 65 and 45 percent quality, and cap their long side at 2500, 1800 and 1200 pixels. A Custom level lets you set the quality, the largest image side and the scan DPI yourself, and a Convert images to grayscale box saves a little more. Text, vector drawings and embedded fonts are left alone at every level, and CMYK, indexed colour and 16-bit images are skipped to avoid colour errors. That is why images are where the real savings come from.

A worked example: two very different PDFs

Same button, opposite outcomes.

Take contract-final.pdf, a 42 page text document exported from a word processor. Lossless trims a little from the structure. Medium does no better, because there are no images to re-encode, and the tool tells you only the file structure was cleaned up. For a file like this, the size you have is close to the size you'll get.

Now take lease-scan.pdf, 12 photographed pages. Lossless barely moves it, because nearly all of its weight is page images. Medium re-encodes those images at a lower quality and resolution, and the new size in the file list shows a large drop. Look at the compressed preview before you press Download: if the fine print is still readable, you're done, and the file saves with a -compressed suffix.

Scanned document mode for files that won't shrink enough

The last resort, with a price.

Some scans store their pages in image formats the levels skip, such as CMYK, and some limits are simply very low. Scanned document mode redraws every page as an image at 150, 120 or 96 DPI, depending on the level, and saves each page as a JPEG in a new PDF. This is usually how a scanned mark sheet or ID gets under a strict limit.

The trade-off is real. Text can no longer be selected or searched, links stop working and form fields are flattened. Used on a text document it can even make the file bigger, and the tool warns you when a result didn't get smaller. Keep this mode for scans, which were pictures of text all along.

Getting under an upload limit, step by step

Stop at the first setting that fits.

Each step costs a little more quality, so work up the list and stop as soon as the new size fits. Or tick Aim for a target size, type the limit, and let the tool run these steps in order and keep the first result that fits.

Start with Medium and read the new size.
Still too big? Try Strong, then tick Convert images to grayscale.
Still too big and it's a scan? Turn on Scanned document mode.
Check the preview each time and make sure the smallest text you care about is still readable.
Don't keep re-compressing the output. Each JPEG pass loses more detail; go back to the original and pick a stronger level instead.

Mistakes that keep a compressed PDF too big

Five patterns behind most failed uploads.

When a compressed file still bounces off the limit, one of these is usually the reason.

Expecting a text file to shrink like a scan. Text and fonts are already compact, so a document with no images changes little at any level.
Compressing an encrypted file. Password-protected PDFs aren't accepted; remove the password with Unlock PDF, compress, then add it back with Password Protect PDF.
Ignoring the page count. A 100 KB limit and a 40 page scan may never meet. Remove pages you don't need or split the file.
Printing to PDF to flatten it first. That often turns text into images and makes the file larger, the opposite of the goal.
Using Scanned document mode on a file someone needs to search or copy from. The text layer is gone afterward.

When Compress PDF is the wrong tool for the job

Sometimes smaller means fewer pages.

Compress PDF works on what's inside each page. When the problem is how many pages there are, use a page tool instead: Split PDF breaks a document into uploads that each fit the cap, Extract PDF Pages pulls out just the section a reviewer needs, and Delete PDF Pages drops the blank scanner pages that add weight for nothing.

It also helps to fix the source. A scan made at a sensible resolution, or photos shrunk before they go into JPG to PDF, starts life smaller and needs less compression afterward.

Try it now

Open Compress PDF

The tool is one click away. No sign up, no upload, no payment.

Open Compress PDF

Write a review