Why compressing a PDF sometimes makes it bigger

Run a scanned contract through a PDF compressor and it might drop from 20 MB to 2 MB. Run a text-only report through the same tool and it can come out larger than it went in. Same tool, same settings, opposite results.

The difference is not the tool. It is what the two PDFs store inside.

A PDF is not one kind of file

The .pdf extension covers two fundamentally different things:

A text PDF — exported from Word, Google Docs, LaTeX, or a website — stores instructions: place this character of this font at this position. A full page of text is a few kilobytes, because “the letter A at position x,y” costs almost nothing to write down. The fonts themselves are embedded once and reused across every page.

A scanned PDF — from a scanner or a phone app — stores photographs. Each page is a large image, typically a JPEG, wrapped in PDF packaging. A single page at 300 DPI is easily 1–2 MB, because a photo of text costs vastly more to store than the text itself.

Most PDFs in the wild are one or the other, and everything about compression follows from which one you have.

Why scans compress so well

A scan from a typical phone app is saved at high resolution and high JPEG quality — settings chosen to be safe, not small. There is usually enormous headroom: the image can be re-rendered at a lower resolution and re-encoded at a lower quality with no visible difference on screen.

When we tested this while building the PDF compressor, a 2.06 MB image-only document came out at 254 KB — 88% smaller — at 96 DPI and 80% JPEG quality, with no visible change at normal reading size. That ratio is typical for scans, not exceptional.

Why text PDFs get bigger

Now run the same process on a text PDF. The compressor renders each page to an image and stores that image. But the original page was a few kilobytes of character positions — and any image of that page, at any reasonable quality, is tens of kilobytes.

In the same test, a three-page text document went from 1.9 KB to about 66 KB — thirty times larger. The tool did exactly what it was asked to do. There was simply nothing to save: text stored as text is already about as small as that page can possibly be.

This is why a good compressor should tell you when the result is larger rather than silently handing you a worse file.

The hidden cost: the text layer

Rasterising compression — the kind that re-renders pages as images — has a second consequence that matters more than file size: the text stops being text.

In the compressed copy you cannot:

• select or copy a sentence
• search for a word with Ctrl+F
• have a screen reader read the document aloud
• let any system extract data from it automatically

The file looks identical and prints identically, but it has become a picture of a document. For anything a person will only read, that is usually fine. For anything that will be processed — a contract, a form, an application an employer or an authority runs through their systems — send the original.

How to tell which kind of PDF you have

Two quick checks:

1. Try to select text. Open the PDF and drag across a sentence. If the text highlights, there is a text layer. If your cursor draws a selection box as if over a photo, it is a scan.
2. Check the size per page. Divide file size by page count. Under ~100 KB per page usually means text; over ~500 KB per page almost always means images.

Or drop it into the PDF to text extractor — if plenty of text comes out, it is a text PDF; if the tool warns you the document looks scanned, it is images.

Practical rules

• Scanned or image-heavy PDF, too big to email? Compress it. Expect 60–90% savings. 96 DPI is enough for screen reading; use 150 if the scan contains small print.
• Text PDF that is somehow huge? The size is usually embedded images or fonts, not the text. Rasterising will hurt more than help — the practical fix is re-exporting from the source document with images downsampled.
• Anything that must stay searchable or machine-readable? Do not use a rasterising compressor at all, and keep the original regardless.

The compressor here states this trade-off above the tool and reports the before-and-after size exactly, so you can judge the result rather than trust it.