Icon Fonts and the "Phone Symbol" Trap
By Umut Barış Terzioğlu · · Updated
A handset in front of the phone number, an envelope in front of the email address, a pin next to the city: icons on a CV save space and look neat. To a parser they are not pictures. Depending on how they were added, they come out of the file as unknown characters, as nothing at all, or as stray symbols glued to the details next to them.
This is the third part of a series on how parsers read CV files. The first part on ATS text extraction explains why only the text layer counts. Here we look at two typography features that damage that layer without changing how the page looks: icon fonts and ligatures.
Why icon fonts turn into unknown characters
Many icon sets, Font Awesome among them, are delivered as fonts. Each icon is a glyph assigned to a code point in the Unicode Private Use Area (U+E000 to U+F8FF), a range the standard deliberately leaves without meaning. On screen, the font draws a phone. In the extracted text there is only a code point that no other font or program knows how to interpret. Text editors usually show it as an empty box or a question mark.
Icons inserted as images or vector shapes behave differently: they leave nothing in the text at all. That is harmless in itself, but if the icon was the only label, the parser has to guess what the number next to it means. Emoji are ordinary Unicode characters, so they survive extraction, but they still tell the parser nothing it can use.
Most parsers recognise an email address or a phone number by its pattern, so a missing label is not the main risk. The risk is clutter: a private-use character glued to the start of a number, or an icon and an address that end up on different lines.
Why ligatures drop letters
A ligature is a single glyph that replaces two or three letters so they look better together, for example fi (U+FB01), fl (U+FB02), ff (U+FB00) and ffi (U+FB03). Whether the letters come back on extraction depends on how the PDF was written. If the file maps the ligature glyph back to its letters, nothing is lost. If that mapping is missing, the pair can come out as one unusual character, a gap or a question mark:
On the page: Profile · Certified · Workflow · Efficiency
Extracted: Pro le · Certi ed · Work ow · E ciency
A search for "certified" or "workflow" then finds nothing, even though both words are clearly visible on the page.
What to change
| Problem | What lands in the text | Fix |
|---|---|---|
| Icon font in front of contact details | Unknown character, often glued to the data | Write "Phone:" and "Email:" as words, or leave the label out |
| Icon as an image or shape | Nothing | Fine on its own, but never let an icon replace a word |
| Skill or language shown only as an icon | Nothing usable | Write the skill or language as text |
| Standard ligatures in the PDF | Missing letters or odd characters | Turn ligatures off before exporting |
In Microsoft Word, ligatures are set in the Font dialog, on the Advanced tab under OpenType features. Design tools and typesetting systems often apply them automatically, so check their typography or export settings. After any change, export again and repeat the copy test below.
Broken accented letters, such as Turkish or German special characters, look like a similar problem but have a different cause; see special characters on a CV.
A quick check for your own file
- Copy all text from the PDF and paste it into a plain text editor.
- Look at the contact lines. Is there an empty box, a question mark or an unknown symbol next to the phone number or email address?
- Search the pasted text for words containing fi, fl or ff, such as "profile", "certified" or "office". Are they complete?
- Paste the text into the quick test. It flags private-use icon characters and leftover ligature glyphs as parseability findings. Letters that have vanished completely only show up when you read the text yourself.
In the full analysis, the parser view lists your name, email, phone, location and links as found, suspect or missing, so you can see whether an icon got in the way. The last part of this series moves from single characters to whole sections and looks at section headings.