How Embedded Fonts Bloat a PDF and How to Fix It with Subsetting
A PDF with one page of plain text has no business being several megabytes — yet it happens constantly. The usual culprit is not the text at all, but the fonts embedded alongside it. Understanding font embedding explains the mystery and points straight at the fix.
The situation
You exported a short document — a cover letter, a certificate, a one-page memo — and the PDF came out at 3, 5, even 8 MB. There are no images. The text would fit in a tiny file. Something invisible is inflating it, and that something is almost always the way fonts were stored inside the document.
Why fonts bloat a PDF (the mechanism)
To guarantee your document looks identical on every device, a PDF can embed the actual font files it uses. That is a good thing: it is why your carefully chosen typeface does not get substituted on someone else's computer. The problem is how much of the font gets embedded.
Full font embedding stores the entire font file — every glyph, every weight, all the kerning and hinting data — even if your document only uses a few dozen characters. A single modern font, especially one with a large character set (think fonts that cover many languages or include thousands of CJK glyphs), can be several megabytes on its own. Embed the regular, bold, and italic styles of two such fonts and you have stacked up 10+ MB before a single word appears.
Subsetting stores only the glyphs the document actually uses. If your memo uses 70 distinct characters, subsetting embeds just those 70, plus the data needed to render them. The visual result is identical, but the embedded font data can shrink dramatically.
So a short text PDF balloons when: multiple full fonts are embedded, each font has a large glyph set, and several styles of each are included. None of that is visible on the page, which is why the size feels inexplicable.
How to fix font bloat
Method 1: Subset fonts on re-export
The cleanest fix is to re-export the PDF with subsetting turned on. Most authoring and PDF tools have a setting like "subset embedded fonts when percent of characters used is less than 100%." Enabling subsetting tells the exporter to embed only the glyphs in use.
Method 2: Re-process with Ghostscript
Ghostscript's pdfwrite device subsets fonts by default when it rewrites a file, which is why running a PDF through it often shrinks font-heavy documents:
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.5 \
-dSubsetFonts=true -dEmbedAllFonts=true \
-o subset.pdf input.pdf
-dSubsetFonts=true keeps only the used glyphs; -dEmbedAllFonts=true keeps the document portable so nothing gets substituted on another machine. You get a smaller file that still displays correctly everywhere.
Method 3: Reduce the number of fonts
Every distinct font and style you use is a separate embedded subset. If a document mixes four typefaces across six styles, consolidating to one or two families removes whole subsets at once. Fewer fonts means fewer embedded files.
Method 4: Re-export to a target size
If you want the file under a specific limit, a target-size step re-writes the document — including subsetting the fonts — to fit. PDF300 lets you enter the target size, so the font and stream cleanup happens as part of hitting your number.
Checklist
- If a text-only PDF is large, suspect fonts before anything else.
- Re-export with font subsetting enabled so only used glyphs are embedded.
- Keep "embed all fonts" on while subsetting, so the file stays portable.
- Reduce the number of distinct typefaces and styles in the document.
- Run the file through Ghostscript's
pdfwriteto subset fonts automatically. - Use a target-size re-export when the file must land under a specific limit.
FAQ
Why is my one-page text PDF several megabytes?
Almost always because entire fonts were embedded rather than subsetted. A full font file can be several megabytes on its own, and multiple fonts and styles stack up. The visible text is tiny; the invisible font data is the weight. Subsetting the fonts on re-export usually fixes it.
Should I just stop embedding fonts to make the file smaller?
No, that risks the document looking wrong on other devices, because the viewer will substitute a different font. The better path is to keep fonts embedded but subsetted, so only the glyphs you use are stored. You keep perfect portability and still cut most of the bloat.
Does subsetting change how the document looks?
No. Subsetting embeds only the glyphs the document actually uses, so every character you typed renders exactly as before. The difference is purely in file size. You only run into trouble if someone later edits the PDF and types a character that was not in the original subset.