Why Font Embedding Matters When Converting Word to PDF

The same file, two different appearances

You prepare a document carefully, align the headings, fit the table neatly onto the page. You send the file, and a screenshot comes back from the other side with the question "does it look like this for you too?" Lines have shifted, the heading has broken across two lines, the table has overflowed.

Behind this situation there is almost always a single mechanism: font substitution. This article explains why substitution happens, how embedding prevents it, and which choices determine the outcome when moving from Word to PDF.

A Word document does not carry its fonts

The key point to understand is this: a .docx file does not carry the fonts it uses inside itself. It carries only their names.

When your document says "Calibri, 11 point", what is stored inside the file is that name and that size. The information about how the letters are actually drawn lives not in the file but in the font file installed on the operating system of the computer opening the document. When Word opens the document, it looks up that name on the system, finds it, and draws the letters accordingly.

There is sound reasoning behind this design choice: fonts are copyrighted software products, and copying them into every document would be both problematic in licensing terms and wasteful in file size. But the choice makes a document's portability directly dependent on the font being available everywhere.

How substitution happens and why it breaks the layout

When the system opening the document cannot find the requested font, it does not raise an error; it quietly uses the closest alternative it can find. This behaviour seems reasonable from a user-experience standpoint — showing something is better than showing a blank page. But it makes the result unpredictable.

The problem is not merely that "the letters look different". The real issue is metrics. In every font, each letter has its own particular width. If the letter "m" is 12 units wide in one font, it may be 13 units in the substituted font. In a single letter this difference is trivial; but a paragraph contains thousands of letters.

As these differences accumulate, the total length of a line changes. When line length changes, the point at which text wraps to the next line changes. When line breaks change, the number of lines a paragraph occupies changes. When the line count changes, page breaks shift. In this way a single unavailable font can break the entire page layout of a thirty-page document through a domino effect.

Tables are especially exposed to this effect. If the text in a cell no longer fits because of the substituted font's wider letters, it wraps to a new line, the cell grows taller, the row grows taller, and the table no longer fits the page.

PDF's solution: embedding

PDF solves this problem by doing what Word does not — it puts the font inside the file.

When embedding is performed, the PDF file carries within itself the glyph definitions needed to draw the text. The program opening the file no longer searches the system; it uses the definitions inside the file directly. As a result the document looks exactly the same even on a computer where the font is not installed at all.

This is the technical foundation of PDF's promise that it "looks the same everywhere". That promise holds only if the fonts really are embedded; a PDF without embedding continues to carry the same fragility as the Word document. Indeed, most of the "it looked broken even as a PDF" cases encountered in practice are PDFs without embedding.

Full embedding and subset embedding

Embedding comes in two common forms, and the difference between them shows up in file size.

Full embedding places the font's entire character set into the file. A font may contain thousands of characters, covering the Latin alphabet, Cyrillic, Greek letters, mathematical symbols and punctuation. Even if your document uses only one percent of them, all of them go into the file.

Subset embedding takes only the characters that actually appear in the document. If you wrote a text in English and never used a Cyrillic letter, Cyrillic characters are not placed in the file. This significantly reduces file size, especially in documents that use many different fonts.

Visually there is no difference between the two — the document looks correct in either case. The only practical drawback of subsetting is that if you later attempt to edit text in that PDF, you may run into trouble when you need to type a character that is not in the file. Since PDF is not a format designed for editing in the first place, this is a theoretical limitation for most users.

So what can go wrong?

To be honest, embedding is not always guaranteed.

Licensing restrictions. Inside font files there is a flag indicating whether that font may be embedded in documents. Some commercial fonts are marked as not permitted for embedding. If you used such a font, embedding is skipped during conversion and substitution takes over. This happens quietly; no warning is shown to you.

Fonts absent from the system. If the conversion is performed not on your computer but on a server, the special font you used in the document may not be installed there either. In that case there is nothing to embed; substitution has already occurred at the conversion stage, and the PDF is fixed in its substituted state.

This second point is important because it is often misunderstood: converting to PDF does not restore a font that has gone missing. If the font is absent at the moment of conversion, the PDF makes that broken appearance permanent. So if you are using a rare font, the safest route is to perform the conversion on your own computer where the font is installed and share the result.

What to do in practice

Prefer widely available fonts. Using fonts found everywhere, such as Arial, Times New Roman or Calibri, in body text largely eliminates the risk of substitution. If corporate identity requires a special font, limiting it to headings reduces the damage.

Check the output on a different device. Everything looks correct on your own computer, because the font is installed there. The real test is opening the PDF on your phone or another computer. If the line breaks and page numbers are identical, embedding has worked.

Treat the increase in file size as normal. Embedding makes the PDF larger. In a document using several fonts, this increase may amount to a few hundred kilobytes. That is a reasonable price to pay for the document looking correct everywhere.

Being aware of this mechanism when converting your Word document to PDF removes the "why is it fine for me but broken for them" question before it can arise.

The situation with archival documents

For documents to be kept for years, embedding becomes not merely a convenience but a requirement. This is exactly why the PDF/A archival profile mandates that fonts be embedded. The logic is simple: a font that is common today may be available on no system twenty years from now. An embedded document remains readable then; a non-embedded one opens substituted and with its layout broken.

For documents that constitute official records, contracts or long-term corporate archives, observing this distinction means solving a problem today that would otherwise be encountered a decade later.

In summary

A Word document does not carry its fonts, only their names; this is why the same file can look different on different computers. PDF removes this dependency by embedding the font inside the file, ensuring the document looks the same everywhere. However, embedding sometimes fails to occur because of licensing restrictions or because the font is absent in the conversion environment. The most robust approach is to use widely available fonts and to check the output once on a device where the font is not installed.