You open your resume, it looks clean, well-organized, professionally designed. An Applicant Tracking System opens the same file and, in the worst cases, gets back something closer to a jumbled word-search than a resume. Both of you are looking at the same document. Only one of you is reading it.
Multi-Column Layouts
A two-column resume, contact info and skills down the left, experience down the right, reads perfectly to a human eye trained to scan left-to-right, top-to-bottom within each column. Text extraction software often doesn't understand columns at all, it commonly reads straight across the page line by line, which means it can interleave your skills list into the middle of your job descriptions in an order that makes no sense.
The fix is a single-column layout. It looks less "designed," but it's the layout that reliably survives extraction intact.
Text Inside Images, Icons, and Logos
A little phone icon next to your number, a LinkedIn logo next to your profile link, a designed header with your name in a graphic. All of that is visually nice and, to most parsers, invisible. Extraction pulls text, not images, so anything rendered as a picture rather than actual characters often just doesn't exist as far as the ATS is concerned.
The fix: plain text for anything you need the system to actually read, including your name and contact details. Save the visual flair for a version of your resume a human will open directly.
Tables
Tables are a natural way to lay out, say, a skills matrix or a project timeline, and a common source of scrambled output. Depending on the parser, a table's cells can get read in an order that has nothing to do with how they're laid out visually, sometimes column by column, sometimes with content merged together in ways that lose all structure.
The fix: use plain bullet points instead of a table for anything that needs to parse cleanly, even if a table would look nicer to a human reader.
Headers and Footers
Contact info tucked into a document header or footer, a common way to keep it visible on every page of a multi-page resume, is frequently skipped entirely by extraction tools, which often only read the main body of the document. Put your name and contact information in the body of the first page instead.
Non-Standard Fonts
A decorative or script font might match your personal brand, but if it's not a font the parsing software recognizes cleanly, individual characters can be misread or dropped. Stick to standard, widely supported fonts. This is one area where "boring" and "reliable" are the same choice.
Scanned Images Instead of a Real Document
A resume exported as a flattened image, or a scan of a printed copy, has no extractable text at all unless the system runs OCR on it, and not every ATS does. Always export a PDF from a real document (Word, Google Docs, LaTeX, or a builder that generates one directly), never a scan or a screenshot saved as a PDF.
How to Actually Check
Short of testing against a specific ATS, the simplest gut check: select all the text in your resume PDF and copy it into a plain text file. If what comes out is a jumbled, out-of-order mess, that's roughly what a lot of ATS parsers are also going to see. If it reads top to bottom in a sensible order, you're in reasonable shape.
The UnlockRole resume builder generates a clean, single-column, parseable layout by default, so this is one thing you don't have to manually audit yourself.