The file size usually comes from one of a few specific things
PDFs don’t have a fixed size for a given amount of content — a 20-page document can be 200 KB or 20 MB depending entirely on how it was created. A few causes explain nearly every oversized PDF:
Every page is actually an image. This is the single biggest cause. A scanned document, or a slide deck / report exported with “flatten to image” style settings, stores each page as a full-resolution photo rather than text and vector graphics. A page of typed text might be a few kilobytes; the same page as a scanned image can be a megabyte or more.
Embedded images weren’t compressed on the way in. A PDF built from a design tool or word processor can embed source-resolution photos — sometimes far higher resolution than the page will ever display them at — without the export step compressing them down first.
Fonts are fully embedded, more than once. Embedding fonts keeps a document looking right on any device, but a PDF assembled from multiple source documents can end up embedding the same font’s full character set several times over.
The internal structure is inefficient. PDFs accumulate unused objects, redundant metadata, and uncompressed internal streams — especially files that have been edited, re-saved, or merged from other PDFs — with no visible effect on how the page looks, just wasted bytes.
What compression can and can’t fix
Images respond best. Photos that went in uncompressed, and scanned pages, get re-encoded as JPEG at the highest quality that fits your target. Inefficient internal structure gets repacked and unused objects are dropped, with no visible change. Duplicated fonts are harder: this tool doesn’t merge or trim fonts, so a file bloated by repeated fonts is better fixed by re-exporting it from the program that made it.
What compression can’t do is turn a scanned image back into real text — if you need a scanned document to be text-searchable, that’s a different problem (optical character recognition), not a file-size problem. And a document that’s already lean — mostly short text, well-structured, no oversized images — simply doesn’t have much room left to shrink; if that’s your file, you’ll see it reach a target almost instantly, or find that a very small target genuinely isn’t achievable.
Finding out which one you have
The fastest way to tell is to try compressing with a loose target first, say half your current file size. If it gets there with no visible change, the file was mostly packaging overhead. If it only gets smaller once the images start to look softer, it’s image-heavy, and that’s the trade-off you’re working with.
Try it
Compress your PDF now and see which category yours falls into from the measured result. If you already know the exact size you need, see how to compress a PDF to a specific size; if compression seems to have stopped working entirely, see why a PDF sometimes won’t compress any smaller.