Short answer: drop the PDF and press “Compress and download”. Pages, text and images are left exactly as they are; the saving comes from the file's structure. The result line shows the real before/after size, so a file that cannot shrink is reported instead of being padded with a fake percentage.
What “lossless” actually means here
Most “compress PDF” pages re-render every page as an image. That always shrinks a file and always damages it: text stops being selectable, thin lines go grey and a printed copy looks soft. This tool takes the other route — it loads the document with a real parser, writes it back with compressed object streams, and removes data that only a viewer uses, such as page thumbnails and XMP metadata. If the result is not smaller, you get the smaller of the two.
When the file will not shrink
A PDF that is already optimised, or one that is a single large scan, has almost nothing to strip. In that case the honest answer is the one shown: a few kilobytes, or none. For scans, the useful lever is resolution and colour depth — that is a different tool with a different trade-off, and it is linked below rather than hidden inside this one.