Use a suitable flat document
Choose an unsigned PDF up to 20 MB and 30 pages. This workflow rejects forms, annotations, links, actions, attachments, layers and accessibility tags rather than risking changes to their behavior. If any of those features matter, keep them in the original and use a suitable document editor instead of stripping them just to apply a mark.
Choose wording and scope
Enter one line of up to 80 characters. Blank page selection marks every page; 1-3,5 marks only those physical PDF pages. Select a restrained gray for a draft or a stronger color when the mark needs to stand out. Fonts come from the browser, so multilingual text and uncommon characters require a careful preview for substituted or missing glyphs.
Balance visibility with readability
Start with diagonal placement, 70% width and 30% opacity. Horizontal placement suits short central labels. The mark is centered within the visible crop box, compensates for standard page rotation and is reduced when necessary to fit. Different page sizes can therefore produce different physical watermark sizes. Keep important content readable.
Inspect a real rendered page
Generate the result, then select Preview first marked page. That preview is rendered from the resulting PDF, not an unrelated mockup. Download and check every selected page in your own PDF reader, including landscape, rotated and cropped pages. Original page content is retained; the added watermark is a transparent image and is not selectable text.
Do not confuse a mark with protection
A PDF editor may remove or cover the watermark. The mark does not encrypt content, prove identity, create a digital certificate or remove hidden data. For confidential text, use a genuine redaction workflow and verify the resulting file. Clearing this tool discards the local inputs and results; it cannot withdraw a copy you have already shared.