Why compressing a PDF doesn't always shrink it much

Run a 40MB PDF through a compressor and get back 38MB, and it feels like the tool didn't do its job. Usually it did — the file just wasn't holding the kind of weight that compression can remove. What's inside a PDF determines how much a compressor can help, and not every PDF is carrying the same kind of bulk.

Where PDF size actually comes from

A PDF file's size is almost never the text. A page of typed text — even a hundred pages of it — is tiny, often just a few kilobytes total, because text is stored as compact character codes and font references, not as pixels. The size in a typical large PDF comes from one of three places:

Embedded images. Scanned documents, photos, screenshots pasted into a report — these are stored as raster image data, and image data is what compression tools actually target. This is where a compressor does its best work.

Embedded fonts. If a document uses a non-standard font, the PDF often has to embed the entire font file (or the subset of characters used) so it displays correctly on a device that doesn't have that font installed. This adds a fixed amount of size that compression barely touches, because font data doesn't compress well — it's already fairly dense.

Structural overhead. PDFs generated by certain tools (some scanners, some "print to PDF" drivers) embed redundant metadata, unused objects, or bloated internal structures. This is compressible, but usually accounts for a small fraction of a large file compared to images.

Why a scanned document compresses dramatically and a typed report doesn't

A scanned page is, underneath the PDF wrapper, just a photograph of a page — every scanner captures it as a high-resolution raster image, often uncompressed or lightly compressed to preserve quality for OCR. Re-encoding those images at a lower quality setting or a more efficient format is where the large size reductions happen: a 20-page scan can genuinely go from 15MB to 2MB, because most of that 15MB was image data with real room to shrink.

A typed Word document exported to PDF, by contrast, has almost no image data to compress. If you run it through a compressor and it barely shrinks, that's not a broken tool — there's simply nothing image-heavy in the file for compression to work on. The size you're seeing is font embedding and structure, not something a "compress" operation is designed to touch.

What actually helps in each case

For scanned or photo-heavy PDFs: lower the image quality setting if your compressor exposes one, and check whether the original scan resolution was higher than you need — a document scanned at 600 DPI for archival purposes is often fine at 150–200 DPI for anything that will only be read on a screen.

For text-heavy PDFs that won't shrink: check whether fonts are being embedded unnecessarily. If the document only uses standard system fonts (Arial, Times New Roman, Calibri), some PDF tools let you skip embedding them, since the reader's device already has them — this can meaningfully cut size on documents that otherwise have almost nothing to compress.

For PDFs that combine both: compress first, then check the result — image-heavy pages will show the reduction; text-only pages won't move much, and that's expected rather than a failure.

Try the PDF Compress tool — it processes the file in your browser, so nothing is uploaded, and you can compare the before/after size immediately.