How to Check If AI Can Read Your Resume

This is the fastest check and it catches the failure that matters most. Open your resume in any PDF viewer, press select-all, copy, and paste into a plain text editor such as TextEdit on macOS or Notepad on Windows.

Read what lands there as if you had never seen the document. Three things tell you the answer:

• Nothing pastes. The file has no text layer. It is a picture of a resume, usually because it was scanned, exported from a design tool as an image, or screenshotted.
• Text pastes but the order is scrambled. Your skills column has been interleaved line by line with your work history. A parser reads it the same way you just did.
• Text pastes cleanly in reading order. Name, then contact details, then sections in sequence. This is the outcome you want.

A missing chunk is worth as much attention as garbled text. If your phone number never appears in the paste, it is sitting in a page header or a text box, and it is invisible to the same software that would use it to contact you.

You may be wondering whether clean text is enough, given that screening increasingly runs on language models rather than keyword parsers. It is not quite enough, and this second test closes the gap: instead of checking that the characters survive, check that the meaning survives.

Paste the extracted text into any chatbot and ask it to fill in a form: current title, employer, start and end dates for each role, total years of experience, and top five skills. Then compare its answers against the truth.

What you learn is specific. If it reports the wrong current employer, your reverse-chronological order is ambiguous. If it cannot produce dates, your date format is inconsistent or buried. If the skills it names are not the ones you want the job to see, they are placed too far down the page.

Take a data analyst applying through a large retailer's careers portal. Her PDF pasted cleanly, so she assumed it was fine. The read-back test returned her second-most-recent employer as current, because she had put a contract role in a differently styled block that read as a sidebar. One formatting change fixed it.

Every result above maps to a concrete cause.

• No text layer. Rebuild the resume in a word processor or a builder and export to PDF directly. Never print to image or scan a printout.
• Interleaved columns. Move to a single-column layout. Two columns look organized to a human and read as alternating fragments to a parser.
• Missing contact details. Take them out of the page header or footer and put them in the body of the first page.
• Missing whole sections. Text boxes, tables, and graphics-embedded text drop out. Use plain paragraphs and standard bullets.
• Odd characters where bullets should be. A decorative font or custom glyph. Switch to a standard font and a standard bullet character.

If rebuilding is the answer, the free builder at /resume-builder exports a single-column PDF with a real text layer by default, which removes the first three failures without you thinking about them.

Passing all three checks means your resume is machine-readable. It does not mean a specific employer's system will handle it identically. Parsers differ, some portals re-process your upload into their own form fields, and a few still ask you to retype everything after the upload.

The tests also say nothing about whether the content is persuasive. Clean extraction gets your words in front of the ranking step; it does not make them the right words. That is a separate problem, and the rest of the resume material sits under /answers/topic/resume.

One habit is worth keeping: re-run the copy-paste test after every edit that touches layout. Changing a font, adding a logo, or nudging a section into a text box can silently undo a resume that read perfectly last month.