About Compress PDF
Most PDFs are bigger than they need to be. The Word export embeds every font even when you used three. The scan was saved at 600 dpi when 200 would have been fine. The phone camera snapped photos at 12 megapixels when the page only needs to be readable. Multiply that across a 40-page document and you've got a 30 MB file where a 4 MB one would have done the same job.
Compress PDF rebuilds the file with smaller image targets, drops duplicate fonts, and strips unused metadata. The output is a real PDF, not an image-only flattening, so the text inside stays selectable and the document remains the same kind of file it always was. Just thinner.
Picking a compression level
There are three presets. Recommended sits in the middle and works for the kind of office documents most people deal with day to day. Strong compresses harder, which is what you want for an archive copy that doesn't need to look great. Light keeps quality high, which is the choice for documents you'll print or hand to a designer.
If you're targeting a specific file size, the size cap option lets you set it. The tool will work toward that ceiling and tell you what it could achieve. Sometimes the document just isn't compressible past a certain point, and that's better to know up front than to find out after three rounds of guessing.
What actually shrinks
Images are the main story. A scan saved at 600 dpi gets downsampled to something closer to what a screen or office printer can render, which is often 150 to 200 dpi. The resampling uses bicubic interpolation, so the result still looks clean at normal viewing sizes. JPEGs inside the PDF get recompressed at a quality level matched to your preset.
Fonts are the second story. Many PDFs embed full font files even when they're only used for a few characters. The compressor subsets the embedded fonts down to the characters that actually appear, which can save a megabyte or two on a document that uses multiple typefaces.
Unused objects, stale metadata, and duplicate resources also go. PDFs accumulate these over edits, and they add up.
When compression earns its keep
Email attachments. Gmail caps at 25 MB, Outlook at 20 MB or 35 MB depending on the plan. A 28 MB report that won't send becomes a 6 MB report that does.
Upload limits on portals. Court filing systems, university submissions, government forms, and a lot of SaaS uploaders refuse anything over 10 MB. Compressing first is faster than discovering the limit when you're already late.
Storage on a phone or tablet. A 200 MB folder of scanned receipts becomes 30 MB after compression, with no visible quality difference at the screen size you actually read them on.
Sharing over chat. Slack, Teams, WhatsApp all have file-size limits or aggressive bandwidth costs. Smaller files arrive faster and don't get stuck in queues.
Limits to expect
Documents that are mostly text don't shrink much. There just isn't enough room in a 40 KB text PDF to find another 30 KB to drop. The savings show up when the document has images, scans, or embedded fonts.
Already-compressed PDFs are a similar story. If a designer ran the file through a strong compressor before sending it, the second pass will only shave a few more percent off, sometimes nothing.
Quality drops are usually invisible on screens but can show up on a sharp print, especially with the Strong preset. If the document is going to a print shop, use Light or skip compression and rely on the print job's own optimizer.
Privacy
The compress pass runs in your browser. The original PDF doesn't leave your machine, and the compressed copy is built right next to it. Closing the tab clears everything from memory.