Someone emailed us last month asking whether our compressor was broken. They had a three page document that came to 14MB, ran it through, and got 13.6MB back. Meanwhile a 300 page novel on their laptop was 2MB and they could not understand how that was possible.
The compressor was not broken. Those two files are not the same kind of thing at all, and once you can tell them apart you stop being surprised.
## Two completely different documents wearing the same extension
A PDF made by a word processor stores your document as instructions. Roughly: put Helvetica 11pt at this coordinate, draw this line, start a new page. The letter A is not a picture of an A, it is a reference to a glyph in a font. That is why a 300 page book is small. Three hundred pages of instructions is not much data, and the font gets stored once no matter how many times you use it.
A PDF made by a scanner or a phone camera stores your document as photographs. Each page is one large image, and the text inside it is not text at all. It is pixels arranged so your eye reads them as letters. A page scanned at 300 DPI in colour is roughly 8.5 million pixels before compression. Three of those and you are at 14MB without anything unusual happening.
Same file extension. Nothing else in common.
## How to tell in about two seconds
Open the PDF and try to select a line of text with your cursor.
If you can highlight individual words, it is a text PDF. If your cursor draws a rectangle over the whole page like it would over a photo, it is a scan.
The other quick check: divide the file size by the page count. Under about 100KB per page and you almost certainly have text. Over 1MB per page and it is a scan. The middle is usually a text document with a few images in it.
## Why compression barely helps a scan
Here is the part that catches people out.
A scan is already compressed. Your scanner saved each page as JPEG before wrapping it in a PDF, so the expensive work has been done once already. Running a generic compressor over it is asking for a second pass of lossy compression on data that has been through it before. You get a few percent smaller and slightly worse looking. That is the 14MB to 13.6MB the person emailed about.
Text PDFs are the opposite. They often carry embedded font subsets, unused metadata, duplicate resources, and revision history from every save. A compressor that strips those can take a real bite out of the file without touching a single glyph, because none of that was ever visible.
So the honest answer is that "compress PDF" means two different operations and only one of them is doing much for you.
## What actually works on a scan
Three things, roughly in order of how much they help.
**Scan at a lower DPI.** 600 DPI is for archiving artwork. For a document that will be read on a screen or printed once, 200 DPI is fine and 150 is often acceptable. Going from 600 to 200 cuts the pixel count by nine. That is the single biggest lever and it costs almost nothing in legibility for ordinary text.
**Scan in greyscale, or black and white.** A colour scan stores three values per pixel. Greyscale stores one. If the original is black ink on white paper, colour is storing nothing except the subtle beige of the paper and the noise in your scanner's sensor.
**Re-encode the images inside the PDF.** This is what a compressor that understands scans actually does. It pulls each page image out, re-encodes it at a lower quality, and puts it back. Our [compressor](/compress-pdf) does this, and it tells you which kind of document it found before it starts, so you know what to expect rather than guessing from the result.
## The one case where nothing will save you
If you scanned a colour photograph, or a page with a large photographic image on it, that data is genuinely irreducible past a point. Photographs are dense. There is no redundancy to remove, which is exactly what makes them look like photographs.
At that point the question stops being how to compress it and becomes whether every page needs to be in the file at all. Splitting out the pages you actually need is usually faster than fighting the ones you do not.
## If a form is rejecting your upload
Check what it is actually asking for before you compress anything. Many portals cap the total PDF at 2MB but also cap each page. A file that passes the total and fails per page needs different handling from one that fails the total, and compressing harder will not fix the first one.
If the limit is genuinely out of reach, scan again at a lower DPI in greyscale. It sounds like more work than running a compressor. It is usually less, and the result is legible instead of smeared.