Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> The problem we have currently is that some CVs fail that review because they’re in a format that the tools cannot parse.

My company (we have software for responding to applicants in real time... think apply... instant conversation with a human recruiter) did a study of about 1,200 resumes and found that 14% of resumes had inaccuracies that would cause them to be screened out... and about 9% of resumes were in a unparsable format. A lot of the screening we see out there is really bad - mostly text search looking for key phrases like specific colleges, specific employers or specific skills. If you are imagining indexing resumes with elastic search and making queries, that may actually be better than state of the art which is usually something that turns into a SQL query.



10 - 15% are typically unparsable, typically PDFs encoded as graphics, and to a lesser extent, Linkedin exports.

14% with incomplete information also sounds right.

And I also concur with you on the screening being pretty basic, not ES level of sophistication at all.


The unparsable numbers are really close to what we see. Lots of work to be done to make hiring better.


Given a lot of these tools cannot even parse CVs with inlined tables, the tools I’m imagining aren’t sophisticated in the slightest. But they are depressingly common.




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: