How an applicant tracking system reads your CV

An applicant tracking system reads a CV by pulling the text out of the file, splitting it into fields such as job titles, dates and skills, then scoring or sorting that text against the words the employer set up for the role. It is a filter before a reader, not a reader itself. A CV that matches the advert's wording in plain, machine-readable text moves forward; a CV that says the same thing in a graphic, a text box or an unusual layout often does not, whatever it actually says.

What does an applicant tracking system actually do with a CV?

Most systems work in three steps. First, the file is parsed: the software extracts raw text from the PDF or Word document and tries to identify sections, contact details, employment history, education, skills. Second, that text is matched against the criteria the recruiter entered, usually keywords from the job advert, sometimes years of experience or a qualification. Third, the system ranks or filters applications so a recruiter sees the closest matches first, or sees only the ones above a threshold.

Acas sets out the wider rule this sits inside: there are no processes required by law for how an employer recruits, but the employer must run a fair process, and discrimination law and data protection law apply throughout, whatever tools are used. An applicant tracking system is a tool inside that process, not a separate legal category, so the same fairness duties apply to the result it produces.

Why do some well-qualified CVs get filtered out?

Parsing fails in predictable ways. A CV built with tables, columns, text boxes or headers and footers often loses its structure when the text is extracted, so a job title ends up detached from its dates, or a skills list disappears entirely. A CV saved as a scanned image or a flattened PDF has no text layer at all, so nothing is extracted. A CV that lists a skill under a different word from the advert, for example "stakeholder management" where the advert says "client liaison", can score low even when the underlying experience matches.

None of this means the content was read and rejected. It means the content was never correctly read. That distinction matters for a candidate deciding how to improve a CV: the fix is usually format and wording, not substance.

Does UK law give you a right to know if a CV was screened by software?

UK GDPR gives individuals rights over a decision that is based solely on automated processing, with no meaningful human involvement, where that decision has a legal or similarly significant effect on the person. The right includes being told about the processing and being able to contest a decision. Most recruitment screening in practice sits below that threshold, because a person still reviews the shortlist before an outcome is decided, but the right exists for the narrower case where a system alone decides an outcome without any human review.

In practice, a candidate who wants to know how a role is screened can ask the employer directly, before or after applying. Acas's fairness duty and the data protection duty both apply to the recruiter regardless of whether a human or a system does the first sort.

How should a CV be formatted so the system can read it?

Use a single column, standard section headings (Experience, Education, Skills), and plain text rather than text boxes, tables or embedded images for anything that should be read. Save the file as a Word document or a text-based PDF, not a scan or a flattened image. Use the same words the advert uses for a skill or a qualification where that word is accurate, rather than a synonym, since the system matches text, not meaning. Spell out an abbreviation at least once, because a system may be set to match either form but not both. Keep job titles and dates on their own line, in a consistent order, so the parser can pair them correctly.

What happens after the system shortlists a CV?

The system's output is a ranked or filtered list, not a decision. A recruiter normally reviews that list, removes anything obviously mismatched, and decides who to invite to interview. The practical test and the bench test that follow an interview, covered elsewhere on this desk, are usually the first point where a candidate's actual work is assessed directly rather than through a parsed CV.

A candidate who is not shortlisted has no way to know, from the outside, whether the cause was the role, the competition, or a parsing problem. Formatting the CV so the text is intact and the wording matches the advert removes one of those causes before the human review stage is ever reached.