Scanners default to settings made for photographs: 600 DPI, full color, and sometimes lossless encoding. A page of black text needs none of that. Re-compressing at a page-appropriate resolution typically removes 80–95% of the file.
1Compress with a target sizeAsk for what you actually need — 2 MB, 5 MB — and let the pages be re-encoded to reach it.
2Compare before and after on the smallest printThe divider under the result shows both versions of the page. If the smallest text is still crisp, the setting is right.
3Re-scan at 300 DPI grayscale if you still have the paperIt is the single biggest saving, and it also scans faster. Use color only when the color carries meaning — a signature, a stamp, a highlighted clause.
4Zip it only for bundlingA ZIP saves little on a scan, because the images are already compressed. It is for sending several files as one attachment.
Why does this happen?
The DPI is far higher than the page needs. 600 DPI is for enlarging and archiving artwork. Documents are legible and print cleanly at 300, and readable on screen at 200.
Color scanning of a black-and-white page. Color triples the data per pixel to record paper that is white and ink that is black.
Lossless or TIFF-style encoding. Some scanners store pages losslessly inside the PDF, which is faithful and enormous. JPEG at quality 75–80 is indistinguishable at document sizes.
The scanner never downsamples. Consumer scanning software often embeds the sensor’s full resolution, whatever the page contains.
Questions
What DPI should I scan documents at?
300 DPI grayscale for anything that will be printed or read closely; 200 DPI is fine for screen-only reading. 600 is for archiving artwork and old photographs.
Will compressing a scan make it unreadable?
Not at sensible settings. The preview exists so you can check the smallest print on your own file rather than trusting a claim.
Why is my scan bigger than the original paper document?
Because a photograph of a page stores every speck of paper texture as data. Text stored as text would be a few kilobytes.