A watermark nobody notices is decoration. One that makes the document unreadable is worse. The gap between them is about fifteen percent opacity.
What is the watermark actually for?
The right settings depend entirely on the job, and there are three distinct ones that get treated as the same thing.
| Purpose | Opacity | Placement | Text |
|---|---|---|---|
| Status marking (DRAFT, VOID) | 15–25% | Diagonal, centre | One short word |
| Ownership and deterrence | 8–15% | Diagonal, centre | Name or domain |
| Recipient tracing | 5–10% | Tiled or corner | Recipient name and date |
Status marks want to be seen — the whole point of DRAFT is that nobody mistakes it for final. Ownership marks want to be present but unobtrusive. Tracing marks want to be nearly invisible while still surviving a screenshot, which is why they are often tiled across the whole page rather than placed once.
Why diagonal beats horizontal
A horizontal watermark across the middle of a page can be cropped away with a single rectangular selection, and it tends to sit in one paragraph and obliterate it.
A diagonal mark crosses the full diagonal of the page, so it touches the top-left and bottom-right corners. Removing it means cropping almost everything. It also distributes itself thinly across many lines rather than concentrating on a few, which is why a diagonal watermark at 15% is more legible to read through than a horizontal one at the same opacity.
Roughly 45 degrees is the convention, and matching the page diagonal exactly — about 55 degrees on A4 portrait — looks deliberate rather than approximate.
Watermarks are not security
This deserves saying plainly, because watermarking is frequently sold as protection and it is not.
A watermark is a layer drawn over the page content. The content underneath is unchanged and fully extractable — the text can be copied, the images can be pulled out, and anyone with a PDF editor can delete the watermark layer in a few clicks. Even without an editor, a screenshot of a lightly marked page followed by OCR recovers the text.
- What it does do: deters casual reuse, marks status unambiguously, makes provenance obvious in a screenshot, and creates a visible signal that a document was not meant to circulate.
- What it does not do: prevent copying, prevent editing, prevent extraction, or survive a determined person with ten minutes.
If the requirement is actually to stop the content being read or copied, watermarking is the wrong tool and no PDF feature will fully deliver it either — permissions and passwords are advisory and widely ignored by non-Adobe readers. The only reliable control over a document is not sending it.
Getting the size and contrast right
Two settings decide whether the mark reads at a glance, and both are usually left at a default that suits neither.
Size. A watermark should span roughly 60–80% of the page width. Much smaller and it reads as a stamp rather than a mark; much larger and the letterforms become so big that individual strokes sit over whole words.
Colour. Mid grey outperforms black at the same opacity, because black at 15% over black text still darkens it noticeably, while grey blends. Red is conventional for VOID and CONFIDENTIAL and should sit at the lower end of the opacity range, since saturated colour draws far more attention than the same value in grey.
Test on the busiest page, not the title page. A watermark that looks fine over a sparse cover can make a dense table unreadable, and the table is the page someone actually needs.
Watermark last, or nearly last
Order matters in a document workflow, and getting it wrong means redoing work.
Watermark after merging, because marking each part separately gives you inconsistent placement across the combined file and no way to fix it short of starting again. Watermark after compressing, because compression re-encodes page images and can leave visible artifacts along the edges of a semi-transparent overlay.
But watermark before signing. A signature covers the file bytes, so adding a watermark afterwards invalidates it.
The sequence that avoids all three problems: merge, compress, watermark, sign. The Watermark PDF tool handles the third step on your device, and Sign PDF the fourth — neither uploads the document, which matters given that watermarked files are usually the confidential ones.
Frequently asked questions
What opacity should a PDF watermark be?
8–20% for most purposes. Below 8% it tends to vanish when printed; above 25% it competes with the text underneath. Status marks like DRAFT sit at the top of that range because they are meant to be noticed; ownership marks sit at the bottom.
Can a watermark be removed?
Yes, easily. It is a layer over unchanged page content, and any PDF editor can delete it in a few clicks. Watermarking deters casual reuse and marks status — it is not a security control and should not be relied on as one.
Should the watermark be diagonal or horizontal?
Diagonal, at roughly 45 degrees. It crosses the full page so it cannot be cropped away with one rectangular selection, and it spreads thinly over many lines instead of obliterating one paragraph — which makes it more legible to read through at the same opacity.
Does watermarking increase file size?
Barely, for a text watermark — it is a few drawing instructions per page. An image watermark adds the image once and references it on every page, so a large logo can add a megabyte or two. Use a small, simple image if size matters.
Should I watermark before or after merging?
After. Watermarking each part separately gives inconsistent placement across the combined document with no easy fix. Merge first, then compress, then watermark, then sign — signing must be last because it covers the file bytes.
Can I watermark only some pages?
Most tools support a page range, which is useful when a cover page or appendix should stay clean. If your tool applies to everything, split out the pages you want marked, watermark them, and merge the document back together.