Like every writer on the planet, I’ve written something about artificial intelligence: the hopeful pros land the devastating cons. For the most part, trying to determine what the future holds amounts to educated guesswork – but in areas where the technology is already in use, speculation gives way to clear, real-world outcomes.
With that in mind, let’s dive into AI in job seeking. What’s the state of play with applicant tracking systems (ATS), the Al-driven screening tools that sort through and evaluate job applications before human eyes ever see them?
With over 98% of Fortune 500 companies using these systems, and in a country where employment, or lack thereof, is a constant and contentious issue, it’s worth dissecting how job seekers can meaningfully navigate the digital gatekeepers of modern hiring.
Across industries, limited vacancies and high application volumes mean more screening. In the South African context, critical skills shortages and the need to meet strict POPIA and B-BBEE requirements often mean recruitment challenges are amplified.
Many ATS systems help maximise efficiency to meet these challenges, with some combination of the following: keyword matching, relevance scoring, formatting interpretation, section identification and file-type compatibility.
Keyword matching
This is fairly straightforward. Depending on the job title, an ATS will pull info from your CV related to job titles, certifications, technical skills (‘Salesforce’ ‘Python’) and soft skills (‘cross-functional collaboration). ATS systems don’t just count keywords; they evaluate context, which is crucial for identifying candidates with practical experience rather than those who simply list skills.
According to Jobscan, a 2025 survey revealed that most recruiters filter by skills (76.4%), followed by education (59.7%) and job title (55.3%). This is a good indication of where to focus your attention when compiling your CV.
Relevance
This goes hand-in-hand with keyword matching. When you submit a job application, the ATS looks through your CV, compares it against the job description and assigns a score. CVs that fall below a threshold are automatically filtered out.
It’s worth noting that this score is not a universal measure of CV quality: it reflects job-specific compatibility, so the same CV may score 40 percentage points differently for identical roles at different companies, simply because of how keywords are prioritised.
Don’t send the same CV to every company. A tailored approach takes more effort but can lead to significantly higher match scores. Systems also check for years of experience and career progression, so ensure that experience claims are substantiated and not merely stated in a skills list.
Formatting
This matters more than most people realise. File format, layout and the use of graphics, headers and footers can all affect how well a system reads your standardisation, it’s best to play by established rules: use fonts like Arial, Calibri or Times New Roman; stick to plain text and simple templates; avoid images, logos, charts and tables entirely.
Section identification
For the same reasons that you’re sticking to widely used fonts, you should aim to use standard section headings such as Work Experience”, ‘Education’ and “Skills”, rather than creative alternatives like ‘What I’ve Done’ or “My Background”. ATS section detection relies on familiar, recognisable labels to locate and score relevant information.
On the subject of what not to do: writing word processing platform’ instead of Microsoft Word’ may cause the system to miss the connection entirely, and your application could be rejected – an easy fix that can make all the difference. Use exact terminology from the job description.
While a two-or multi-column layout might look polished to the human eye, some ATS systems struggle to parse them, reading adjacent columns as a single stream of text and producing nonsensical output that can’t be scored.
Avoid keyword stuffing – overloading your CV with specific keywords or phrases in an attempt to manipulate a ranking system. And because most systems skip headers and footers entirely, make sure your contact info appears in the body of the CV.
Navigating the current job market can feel daunting, but the mechanics of these Al systems, once understood, are surprisingly gameable. That said, the old-school fundamentals still apply: a strong, honest CV tailored to the position advertised. But now that CV just needs to speak a language machine can read.
Get that right, and you give yourself a fair shot at the thing that matters most: getting in front of an actual person.
Words: Tshiamo Seape
Photographs: Gallo/Getty Images
