Choose your PDF on the compress page. It is processed in your browser and never uploaded.
Start with the recommended level. If it is still too big, try the strongest one — then zoom in and check small text.
Compare the before/after size shown on the page. If it is still over the limit, splitting the file in two usually beats compressing harder, because splitting is lossless.
Three things drive PDF size: the resolution of embedded images, whether fonts are fully embedded, and duplicated objects. One A4 page scanned at 300dpi in colour as PNG can hit 3–5MB; the same page as a 150dpi JPEG is often 300–500KB. That is why compression helps scans enormously and does almost nothing for text-only PDFs — the text itself is a few tens of kilobytes, the rest is font data.
Compression does two things: downsamples images and re-encodes them. The text layer, vector graphics, bookmarks and form fields are untouched, so the result is still selectable and still searchable. The real risk is fine detail inside images — small amounts on an invoice, handwritten signatures, the edge of a stamp. Zoom in and check before you send.
The number comes from whoever receives the file, not from PDF itself: email attachments are usually capped at 20–25MB, government and school portals at 2MB or 5MB. Confirm the actual cap first. If 2MB is accepted there is no reason to squeeze down to 1MB and lose detail.
It is already tiny — mostly embedded fonts. Such files rarely hit a size cap; the real blocker is often a page-count limit instead.
No. Compression is lossy; detail that is gone stays gone. Keep the original and redo it at a lighter level.
Yes. Only images are re-encoded; the text layer is preserved.
No. Compression runs entirely in the browser and works offline.