Why Your PDF Looks Different on Another Computer: Font Embedding Explained
The PDF format promises that a document looks identical everywhere, yet everyone has seen the promise break: a proposal opens on a client's laptop with different letterforms, lines wrapping in new places, a table pushed onto the next page, or Chinese characters replaced by hollow rectangles. The root cause is almost always the same. A PDF stores text as character codes plus the name of the font that should draw them. If the font program itself travels inside the file, every machine draws identical glyphs. If only the name travels, the viewer must find that font on the local system — and when it cannot, the trouble starts.
When a named font is missing, the viewer substitutes the closest thing it has. Substitution changes glyph widths, and widths are the dominoes: text reflows, line breaks move, headings wrap, and carefully aligned tables shear. For CJK text the failure is harsher — a Western machine usually has no font covering the needed Chinese characters at all, so the viewer renders the notorious empty boxes known as tofu. The document did not change; the file simply never contained the shapes, only references to shapes it hoped the destination machine would own.
Checking a file takes thirty seconds. Open the document properties in your PDF viewer and find the Fonts tab: each font should be marked as an embedded subset. Subsetting means the file carries only the glyphs actually used — the reason embedding a ten-megabyte CJK typeface does not produce a ten-megabyte PDF. Any font listed without an embedded mark is a substitution waiting to happen on someone else's computer. This check belongs in your pre-flight routine for anything that must survive travel: contracts, resumes, print-shop artwork, anything with non-Latin text.
Prevention happens at export time, because that is when embedding is decided. Word's own save-as-PDF path generally embeds fonts, but its options include a compatibility mode that can skip certain licensed typefaces, and third-party printer drivers are less predictable. PdfWill's Word to PDF conversion produces output with fonts embedded, entirely in your browser with nothing uploaded. Going the other way, PDF to Word rebuilds an editable document from a PDF, though if the source lacked embedded fonts the reconstruction inherits the same substitutions — no converter can recover shapes that were never in the file.
When you receive a broken PDF and cannot get a re-export from the author, your options are honest but limited. If the file misbehaves because its internals are damaged rather than merely unembedded, repair can rebuild the structure. For a document that must display exactly right on one occasion — a slide shown on a borrowed machine, artwork for a print shop — converting pages with PDF to Image freezes the rendering into pixels that no font substitution can touch, at the cost of selectable text. But the durable rule sits upstream: embed at export, verify in document properties, and the file you send is the file they see.