Open the PDF and drag across a sentence: text highlights if it is a text PDF, and the whole page turns blue if it is a scan.
A few hundred KB per page means text; 2 MB or more per page means scan. Text PDFs need a different approach, not more compression.
Scan: compress and convert to greyscale. Text PDF: re-export from the source, or split and send only what is needed.
A PDF generated from Word or Excel stores characters and vector lines — the whole file is often a few hundred KB. Compression works by re-encoding images, and there are no images to re-encode. A 5% drop is the honest result, not a broken tool.
Lossy compression gives its biggest win on the first pass. Running it again downsamples an already-downsampled image: 5-10% smaller, visibly softer text. If you must go smaller, change the output rather than compressing again — lower DPI or drop to greyscale.
Convert colour scans to greyscale (routinely 40-60% smaller), split out only the pages that must be sent, or re-export from the original Word/Excel source, which beats squeezing a scan every time.
Theirs is a scan — full-page bitmaps. Yours is generated electronically and already close to its floor.
Yes for scans: compression lowers image resolution. To keep text crisp, switch to greyscale instead of downsampling further.
It exists, but typically saves 5-15%. Large reductions are always lossy.