A "sign this PDF" tool looks simple: you draw or type your name, click where it goes, and download. Underneath, there are a few different things that can happen. Here is how it works in practice.
Step 1: Turn the signature into an image
A drawn signature is a canvas converted to a transparent PNG. A typed one is text rendered in a script font, either as an image or as real text. Either way the result is just graphics, plus optional text like a date.
Step 2: Stamp it onto a page
A PDF page is a content stream of drawing operators. To add a signature you add the image as an XObject resource on the page, then append a few operators that position and scale it, roughly "save state, move and scale, draw image, restore state". The coordinates are in points (1/72 inch) from the bottom-left of the page, so a tool must convert from where you clicked on a preview (top-left, in pixels) to PDF coordinates. It must also account for page rotation and cropped pages, which is a common source of signatures landing in the wrong place.
Incremental update vs flattening
There are two ways to save the result:
- Incremental update: the new objects are appended to the end of the file, with a new cross-reference section. The original bytes stay untouched. This is cheap and preserves earlier content, but the previous version is still inside the file.
- Full rewrite (flattening): the document is re-serialised with the signature drawn into the page content. The file is cleaner, but the signature can no longer be moved as a separate object, and rewriting can change the file in ways that break any existing digital signatures.
Neither is "more secure" by itself. They are just different ways of saving.
Visual e-signature vs digital signature
This is the part that gets confused most often. A visual signature is a picture of your signature on the page. It shows intent, but anyone can copy the image, and nothing in the file proves who placed it or whether the document changed afterwards.
A certificate-based digital signature (for example PAdES) computes a hash of the document, signs it with a private key tied to a certificate, and stores the result in a signature field. Viewers can then tell whether the file was modified and who signed. Some contracts and jurisdictions require this. Stamping an image does not give you it. Check what your recipient actually needs before choosing.
Why locked PDFs should be refused
PDFs can carry permission flags (no editing, no annotating, no form filling) enforced through encryption. A tool that is meant to respect those restrictions should refuse such files rather than strip the protection. The honest path is to ask the owner for an unrestricted copy, or the owner password.
Practical tips
- Put the signature on the correct page, and use the page list option for multi-page documents (for example initials on every page, full signature on the last).
- Zoom in on the output and confirm the signature does not cover text or a signature line.
- Check that the file opens in at least two viewers.
- Keep the original. Never overwrite it with the signed copy.
- Remember that image signatures with sensitive handwriting should be shared carefully.
If you want to try a browser-based version of this workflow, I built one: https://gotoolsonline.com/tools/pdf-add-signature
Disclosure: I run GoToolsOnline.
