How editing a PDF actually works

Last reviewed on 6 October 2026

People expect a PDF to behave like a document in a word processor: click, type, and the paragraph rearranges itself. It does not, and the reason is not a weak editor. It is what a PDF is. This page explains what is inside a PDF page, what this editor changes when you edit text, and which results you should expect. Knowing this saves time, and it prevents one mistake that can be costly: believing that covered text is gone.

1. A PDF page is a list of drawing instructions

A word processor stores text as paragraphs and works out the line breaks every time you open the file. A PDF stores the finished result. Each page contains a sequence of instructions for a printer or screen: select this font at this size, move to this position, draw these characters; move, draw the next piece; draw this line; place this image.

Simplified, one line of an invoice looks like this inside the file:

Three consequences follow from this, and they explain almost everything an editor can and cannot do:

• Nothing reflows. Line two does not know that line one exists. If line one gets longer, line two stays where it is.
• Text can be chopped up arbitrarily. The program that created the PDF may have written a line as one piece, as one piece per word, or as one piece per change of font. What looks like a single sentence can be several separate instructions.
• The font travels with the file, but often only in part. More about this in section 3.

2. What happens when you edit a line

When you click text in this editor, your browser shows an input box over that line. When you press Enter, the following happens on our server:

1. The PDF engine locates the drawing instructions that produce the line you clicked.
2. It compares the old and the new wording and changes only the part that differs. If you correct one word in a line that also contains a bold word, the bold word is left untouched.
3. It replaces the characters in the page content. If the new text is longer or shorter, the rest of that line is moved along so that nothing overlaps.
4. It saves the result as a new version of the file and sends the changed page back to your browser.

The old wording is not hidden. It is no longer in the page content. That is the difference to the common trick of painting a white box over old text and typing on top, which leaves the original words in the file, where a search or copy-and-paste still finds them.

3. Why the font sometimes changes

To keep files small, most programs embed only the characters a document actually uses. A PDF that contains the word “Invoice” in a certain font carries the shapes for I, n, v, o, i, c and e in that font, and nothing else. This is called a subset font.

If you change “Invoice” to “Voice”, every letter you need is there and the result is indistinguishable from the original. If you change it to “Quote”, the letters Q, u and t are missing. No editor can invent them. The engine then uses a similar font for that piece of text, and this editor tells you with a message when it had to do so.

• Changing digits is usually unproblematic, because documents with numbers tend to contain all ten digits.
• Short headings in a decorative font are the most likely to be affected, because few different letters were embedded.
• If an exact match matters, check the edited line at high zoom before you send the file on.

4. Changing the page versus placing things on top

The tools in the editor fall into two groups, and it helps to know which is which.

Text you add yourself is real text as well: it can be selected and searched in the downloaded file. It is set in one of three fonts that are metrically compatible with Arial, Times New Roman and Courier, and the font is embedded so the result looks the same on every device.

5. Scanned documents

A scanner produces a photograph of a page. Wrapped into a PDF, the page then consists of one big image. There are no characters to find and therefore nothing to click. The editor’s status line tells you when a page contains no editable text.

You can still work with a scan: cover a passage with Whiteout and type the new wording with Add text, fill in a paper form with text and check marks, sign it, and rearrange pages. Turning the picture back into text is a different job called OCR, which this editor does not do.

Some scans carry an invisible text layer that an OCR program added. Editing that layer would change the hidden text but not the visible picture, so treat such files like scans.

6. Whiteout is not redaction

If text must not be recoverable, remove the text itself: click it with the Edit tool and use the delete button, or replace it. That takes the characters out of the page content. Then check the result yourself. Open the downloaded file, search for the removed words, and try to select the area.

Two further points for sensitive documents. Pictures cannot be removed with this editor, only covered, so a photographed signature or ID stays in the file under a whiteout box. And information can live outside the visible page, for example in the document properties (author, title) or in attachments. This editor does not clean those. For documents where disclosure has legal consequences, use a dedicated redaction tool and have the result checked.

7. Forms

There are two kinds of PDF form. Fillable forms contain real fields: the file says “here is a text field called last_name”. The editor shows these as input boxes, and your entries are written into the fields when you download, so the form remains fillable afterwards.

Flat forms are just lines and labels drawn on a page, typically a scanned paper form or a form exported without fields. Here you place text with Add text, tick boxes with the check mark tool, and sign with the Sign tool.

The signature you place is an image of your signature. It is not a certificate-based digital signature that proves who signed or that the file was not changed afterwards. Whether an image signature is sufficient depends on what the recipient or the law requires.

8. What you get when you download

Until you download, the things you placed on top stay movable objects in your browser. On download, the server writes them into the pages of the current version: text as text, rectangles and highlights as vector graphics, images and drawings as pictures. The file you receive is an ordinary PDF that opens in any reader.

Every text edit and page operation creates a new version on the server while the earlier ones are kept, which is how undo works without a limit. All versions belong to your session and are deleted together, 2 hours after your last activity or immediately when you choose “Delete my files now”. The details are on the page Your files and privacy.

9. Limits in one list

• Text is edited one line at a time. Paragraphs do not re-wrap.
• Characters that the embedded font lacks are set in a similar font.
• Scanned pages have no editable text, and there is no OCR.
• Existing images and drawings can be covered but not changed or removed.
• Text that is rotated or set along a curve may not be offered for editing.
• Whiteout covers, it does not remove.
• Password-protected files can be opened only with their password.
• The maximum file size is 1024 MB.
• Document properties, bookmarks and attachments are kept as they are and cannot be edited here.