How to change one word in a PDF without wrecking the formatting

You need to change one date on an invoice. You open a PDF editor, retype it, and the new text comes out in a slightly different typeface, a little too heavy, sitting a fraction too low. The document now looks tampered with, which is precisely the impression you were trying to avoid.

This happens for a specific and fixable reason.

Why the font changes

PDFs usually carry their fonts inside them. That is what makes the format portable: the file does not rely on your computer having the right typeface installed, because a copy travels with the document.

But most PDFs do not embed the *whole* font. To keep files small, they embed a subset containing only the characters actually used on the page. If a document contains the word "March" and nothing else in that font, the embedded subset may hold only M, a, r, c and h. Change it to "April" and the letters p, i and l are simply not there.

Faced with missing characters, most editors give up on the original and substitute something approximate — usually Helvetica or Arial. That is the mismatch you are looking at. It is not carelessness; the tool genuinely does not have the letters it needs.

What editing in place does instead

PDF Clay reads the actual font resource from the document and hands it to your browser, so the editable text you type into is rendered with the document’s own typeface rather than a stand-in. When you export, the replacement is written using that same embedded font.

The tricky part is the fonts with a subset that genuinely lacks your characters. The honest answer is that no tool can conjure a glyph that is not in the file. What can be done is to keep everything else identical — size, weight, colour, baseline and spacing — so a substituted character is as close to invisible as possible, rather than obviously different.

Getting the position right

Matching the typeface is only half of it. Text also has to sit in exactly the right place, and PDFs position text by its baseline — the invisible line letters sit on, with descenders like g and y hanging below. Line up two pieces of text by their tops or their boxes and they will look subtly wrong even in the same font.

The editor anchors every editable region to the original span’s baseline, so your replacement sits exactly where the old text sat. It also measures the width of what you type against the width of what was there, and adjusts letter spacing to compensate for the small differences in how a browser and a PDF renderer lay out the same characters. Without that correction, a long line drifts noticeably by its end.

How to change text

1. 1Upload the PDF. Each page is rendered exactly as it is, with an invisible editable copy of every piece of text laid over the top.
2. 2Click the word you want to change. A caret appears in it.
3. 3Type. The original characters are removed from the file and yours are drawn in their place.
4. 4Press Tab to move to the next piece of text, or click anywhere else on the page.
5. 5Download. The whole set of changes is replayed against the untouched original, so repeated edits never degrade the document.

When it will not match perfectly

• The embedded subset lacks your characters, as described above.
• The page is a scan, so the "text" is part of an image. Run recognition first.
• The text is inside a vector logo or drawing rather than being text at all.
• The document is form fields rather than page content, which behave differently.