Every graphic on your resume parses as nothing. A phone icon instead of the word phone, star ratings for languages, skill bars at eighty percent — parsers extract text, and graphics contain none. The information those elements carry simply vanishes from your database record, and skill bars manage to be misleading to humans too.
Text extraction pulls characters from your file. An icon is a small image or a glyph from a symbol font — either way, no characters. When you replace the label "Email:" with an envelope icon, the parser is left with a bare string floating near nothing; a well-behaved one still pattern-matches the address itself, but you are now relying on luck instead of structure.
Skill bars are worse because they carry the entire data payload graphically. "Python" followed by a bar filled four-fifths of the way parses as, at best, the lone word Python — the proficiency claim evaporates. Five stars for Spanish parses as nothing at all if the stars are images, or as garbage characters if they are symbol-font glyphs. Everything you encoded visually is gone from the record recruiters search.
Even for human readers, skill bars answer a question nobody asked while raising ones you do not want. Eighty percent of what? Self-assessed against whom? Hiring managers routinely mock skill-bar resumes precisely because the format projects confidence about an unmeasurable number. Meanwhile the honest version — plain words plus evidence in your bullets — is what they wanted anyway.
Icons are more defensible aesthetically, but audit what they cost. A LinkedIn glyph without the URL text gives a parser nothing to store. Section headings replaced by symbols defeat the segmenter that routes content into fields. The rule is simple: decoration may accompany information, never replace it. If deleting every image from your resume would lose actual content, that content is currently invisible to machines.
For contact details: plain labeled text — the words themselves, including the full LinkedIn URL as text. For languages: a text scale with recognized levels, like Spanish, professional working proficiency, or a CEFR grade. For skills: a plain list of nouns, with depth shown where it belongs — in bullets that say what you built and what happened, with numbers.
If you want visual polish, spend it where parsers do not look for content: tasteful color on headings, generous whitespace, a strong typographic hierarchy. Those read as design maturity and cost you nothing in the database. Keep one visually rich version for a portfolio site or in-person handoff if your field rewards it, but the file you upload to application portals should carry every fact as selectable text.
Mildly. Most parsers pattern-match emails and phone numbers regardless of labels, so the details usually survive. The risk concentrates in things without recognizable patterns — profile URLs, section headings, proficiency levels — where an icon replaces the only text a parser could have used.
Use recognized text levels: native, fluent, professional working proficiency, conversational — or CEFR grades like C1. These parse as text, mean something consistent to recruiters, and are searchable, which star graphics never are.
Yes — a two-file strategy works well. Upload the plain single-column version to application portals, and bring or link the designed version where a human chooses to look at it: portfolios, in-person interviews, or fields like design where the artifact itself is evidence.
Run your resume through the free scorer to see exactly which of your details survive parsing as text — and which are currently decorating a database record with blank fields.
