Two-column resumes and ATS: what breaks, what does not

What the vendor measured

Textkernel, the parser under iCIMS, SAP SuccessFactors and Lever, published a study on column resumes. Read naively, left to right, a two-column page interleaves: a line from the sidebar, then a line from the experience, then the sidebar again. After a machine-learning fix, the share of column CVs rendering correctly rose from 62% to 90%.¹ That is a real improvement and still one file in ten.

What interleaving looks like

Here is the naive reading order of a two-column fresher resume we built for testing:

Date of Birth: 14/08/2002 my skills and knowledge to grow along with the organisation.
Father’s Name: Rajesh Sharma
Marital Status: Single ACADEMIC QUALIFICATIONS

The parser now has a line that begins with a date of birth and ends with the objective, and a marital status that runs into a section heading.

When a sidebar is safe

Skills, links and languages in a narrow column parse fine even when interleaved, because each is a short, self-contained term. Experience does not: a job title on one line and its company two lines later, with sidebar text between them, is how jobs get merged or split.

The other things that travel with two columns

Templates that use columns usually also use a coloured header band, icon fonts for phone and email, and skill bars. Each is a separate hazard: light text on a band is dropped by some parsers as decoration, icon glyphs come out as private-use characters, and bars carry no parseable proficiency.

1. Textkernel, “Improving extraction from column resumes” (engineering blog). Greenhouse, “Unsuccessful resume parse” help article, lists “columned designs” among parse-failure causes. ↩