What the available data says about automatic rejection
What does happen is a different problem, and a harder one to notice: misparsing, not rejection.
How an automated system actually reads your CV
An ATS doesn't see your CV's visual design. It turns it into a string of plain text, usually top to bottom and left to right, and tries to store that information as structured data: name, experience, skills, each in its own field.
Any formatting element that breaks that linear order can cause a problem: if you use floating text boxes, tables, or a sidebar saved as an image, your information may end up scrambled or incomplete in the system. That isn't a rejection, but a recruiter who later searches that database by keyword may not find you, because your data ended up in the wrong place. It's a findability problem, not a rejection problem.
More than “two columns,” it usually comes down to two specific situations
- Text turned into an image. A sidebar with your photo, contact details, or skills, exported as an image, usually can't be read as text: there's no text there to extract.
- Tables. A table can mix up the reading order of its cells. What looks clear to you in columns may reach the system jumbled, with data from one cell stuck to data from another.