Why a scan will not compress below 100 KB

There are only two kinds for this purpose, and they behave completely differently. A document that began in a word processor is text, vector lines and a few embedded fonts. A document that came off a scanner or a phone camera is photographs of paper. The first is usually already small and compresses further with almost no loss. The second is nearly all image data, and image data is the file.

You can tell which you have in a second: open the PDF and try to select a sentence with your cursor. If the text highlights word by word, it is a text document. If nothing highlights, or the whole page highlights as one block, it is a scan.

A ten-page text document under 100 KB is routine. A single colour scan at 300 DPI under 100 KB is a serious squeeze, and at 600 DPI it is usually impossible without visibly damaging the page.

Rebuilding the file structure removes duplicated resources, unused objects and the redundancy that accumulates when a file has been edited or merged. This costs nothing visually — the text stays selectable, the fonts stay embedded, the layout is untouched.

That alone frequently takes a text PDF well under 100 KB. If it does not, the usual culprit is a logo, a signature image or a chart that was placed at far higher resolution than the page needs.

A page scanned at 600 DPI in colour is roughly 35 megapixels of image before any compression. At 300 DPI it is about 8.7 megapixels — a quarter of the data for a page that still prints and reads perfectly well. At 200 DPI, which is entirely legible for printed text, it is under 4 megapixels.

This is why the most effective thing you can do to a scan is usually not to compress it but to scan it again. Most scanner software and every phone scanning app lets you choose the resolution and the colour mode. A 200 DPI greyscale scan of a printed page is a fraction of the size of a 600 DPI colour one and loses nothing you were going to read.

Greyscale is the second lever and it is a large one. A colour scan stores three values per pixel where greyscale stores one. If the original document is black ink on white paper — which most forms and certificates are — colour is storing nothing but the paper's tint and the scanner's noise.

Portals usually want a specific document, not everything you have. Three pages of a contract at readable quality beats forty pages compressed until the small print is a smear — and the pages you did not send were never disclosed at all, which matters more than the file size.

Extract what is actually required first, then compress what is left. Compressing a hundred pages and then deleting ninety means almost all of that work was spent on pages you threw away.

Any tool that promises to hit an arbitrary byte target on any document is either flattening your pages into low-quality images — destroying the text layer, so the document can no longer be searched, selected or read by a screen reader — or it is going to fail and tell you afterwards.

There is a difference between a tool that promises a number and one that searches for it. Compress PDF to 100KB here takes your target and tries a range of settings against it, keeping the mildest one that fits — and when nothing fits, it says so and names the reason rather than handing back a flattened document and calling it a success. Treat the promise as the warning sign, not the target box: what matters is whether a tool will admit it fell short.

If a portal rejects your file and you have already scanned at 200 DPI greyscale and sent only the required pages, the document genuinely may not fit. At that point the right move is to ask the portal, not to keep compressing.