Compress PDF

Because a browser cannot reach inside a PDF and re-encode the images buried in it, which is what a desktop tool like Ghostscript does. What a browser can do is render each page and save the result as a picture, which is what happens here. That shrinks a scan enormously, because a scan was already a picture. It also flattens any real text into pixels, which is why the panel says so before you press the button rather than after.

Searching, copying and screen-reader access, all of which depend on the text being text. Links stop working, form fields stop being fillable, and any bookmarks or outline structure goes. For a scanned receipt none of that existed anyway. For a contract or a report it is a serious loss, and the honest alternative is to reduce the file another way: split out the pages you actually need, or go back to whatever produced it and export at a lower image resolution.

Not in a browser, and it is better to say so than to pretend. Properly shrinking a text PDF means subsetting the embedded fonts and recompressing the images inside it while leaving the text objects alone, which needs a real PDF library working on the file's internal structure. Ghostscript does it locally and well. If the file is large because somebody dropped in photographs at print resolution, exporting again from the source document is usually faster than any of this.

Because it was already efficient and the tool had nothing to win. A PDF that is mostly text is tiny to begin with - the text is stored as characters and instructions, not as a picture of a page - so turning those pages into images can only add bytes. When the result would be larger, the original is handed straight back rather than replaced, and the tile says the size did not change. If a text PDF is genuinely too large, the cause is nearly always an embedded font or a scanned page hiding among the real ones.