This Happens Every Day
A software engineer with six years of production experience writes a clean, honest CV. They list the projects they shipped, the teams they led, the systems they built. They apply to a role they are genuinely qualified for.
They never hear back.
Not because a recruiter reviewed their CV and decided they were not a fit. No recruiter ever saw it. An applicant tracking system scanned their CV in under a second, compared it against a list of keywords, and decided it did not match. The CV went into a digital discard pile alongside hundreds of others.
This is not an edge case. According to Harvard Business School research, 75% of CVs submitted through online applications are rejected by ATS software before a human being ever reads them. That is three out of every four applicants filtered out by a machine.
75%
CVs Rejected by ATS
Before any human reviews them
88%
Qualified Candidates Filtered Out
Of rejected applicants meet basic qualifications
<1s
Average Time ATS Spends per CV
Keyword scan, not comprehension
98%
UK Employers Using ATS
Of large UK employers use automated screening
How ATS Keyword Matching Actually Works
Most applicant tracking systems work on a simple principle: does the text in this CV contain the same words as the text in this job description? If the overlap is high enough, the CV passes through. If it is not, the CV is rejected.
This sounds reasonable until you think about how people actually write about their work. A developer who has spent three years building React applications might describe themselves as a frontend engineer. They might list Next.js, TypeScript, and component architecture as their core skills. They might talk about performance optimisation and accessibility improvements. All of which demonstrate deep React expertise.
But if the job description says "React developer" and the CV does not contain the word "React" in the right place, the ATS scores it as a poor match. The system cannot understand that frontend engineering with Next.js implies React expertise. It can only count words.
The same problem hits in reverse. Someone who has copied a list of trending keywords into their CV, including React, even though they completed one tutorial two years ago, passes the ATS filter with a high score. The system cannot distinguish between genuine depth and surface-level mention.
Where ATS Keyword Matching Fails: Match Accuracy by Scenario
- ATS Keyword Match (%)
- Skills Weighted Match (%)
Source: SHRM, LinkedIn, Korn Ferry, Indeed, Robert Half, LSE, PwC, Statista (2026)
The Three Ways Good Candidates Get Caught
Automated screening can filter out qualified candidates in three common ways.
Pattern 1: The Vocabulary Gap
The candidate uses different terminology than the job description. They write 'cloud infrastructure' instead of 'AWS'. They say 'agile delivery' instead of 'Scrum'. They describe 'data pipelines' instead of 'ETL processes'. The skills are identical. The words are not.
This hits career changers hardest. A project manager moving into product management has transferable skills in stakeholder management, roadmapping, and delivery. But if their CV uses project management vocabulary and the job description uses product management vocabulary, the ATS sees no overlap.
International candidates face this too. Terminology differs between UK, US, and European markets. A 'software developer' in one market is a 'software engineer' in another. Some ATS systems treat these as different roles entirely.
Pattern 2: The Format Penalty
Many ATS systems struggle with CV formatting. Tables, columns, headers in images, and non-standard layouts confuse the text extraction engine. The system either misreads the content or skips sections entirely.
A beautifully designed CV with skills listed in a sidebar table might have those skills completely invisible to the ATS. The candidate thinks they have highlighted their strengths. The system never sees them.
PDF parsing is particularly inconsistent. The same CV saved as a PDF from different applications can produce completely different text extraction results. Some ATS vendors handle this well. Many do not.
Pattern 3: The Depth Blindness
Keyword matching cannot assess depth of experience. Someone who mentions Python once scores the same as someone who has led Python teams for eight years. The word is either present or absent. There is no middle ground.
This creates a perverse incentive. Candidates learn to stuff their CVs with keywords, listing every technology they have ever touched regardless of their actual proficiency. The system rewards breadth over depth, which is the exact opposite of what most hiring managers want.
Senior professionals are often the worst affected. They tend to write concise, achievement-focused CVs rather than keyword-heavy ones. Their CVs read well to humans but score poorly with machines.
What This Costs Employers
The conversation about ATS filtering usually focuses on candidates. But employers pay a steep price too.
When your ATS rejects qualified candidates, you are not just being unfair to applicants. You are shrinking your own talent pool. You are paying for job advertising that drives applications you never see. You are extending your time to hire because strong candidates get filtered out in the first pass. And you are losing those candidates to competitors whose systems are better at recognising genuine fit.
In the UK IT market specifically, where specialist roles already face a 54% fill rate difficulty, every incorrectly rejected candidate represents real cost. If your ATS is rejecting 75% of applicants and a significant portion of those are qualified, you are making an already difficult hiring market harder for yourself.
The Hidden Cost of ATS False Negatives for UK IT Employers
- Skills Weighted Matching
- Traditional ATS
Source: SHRM, LinkedIn, Korn Ferry, Indeed, Robert Half, LSE, PwC, Statista (2026)
Skills Weighted Matching: A Different Approach
The alternative to keyword matching is not to abandon automation. Manually reviewing every CV is not realistic when a single job posting can generate hundreds of applications. The alternative is smarter automation.
At HirebyAI, our matching system works in two stages. First, we extract structured skills data from every CV using AI that understands context, not just keywords. When someone describes building microservices with Node.js and deploying to AWS, the system recognises backend engineering, cloud infrastructure, and distributed systems expertise, even if the candidate never uses those exact phrases.
Second, we match those extracted skills against job requirements using fuzzy matching that handles synonyms, abbreviations, and related technologies. React and ReactJS are treated as the same skill. Kubernetes and K8s are equivalent. A candidate listing Next.js gets credit for React ecosystem experience.
The result is a score weighted 60% on skills match, with experience alignment and profile fit making up the remaining 40%. Crucially, the score is transparent. Candidates see exactly which skills matched, which were missing, and where they can improve. No black box.
"The best recruitment technology does not just filter faster. It filters smarter. The goal is not to process more CVs per second. It is to make sure the right ones reach a human."
HirebyAI Engineering Team
What Candidates Can Do Right Now
While the industry works on better matching technology, candidates can take practical steps to improve their chances with existing ATS systems.
First, read the job description carefully and use the same language in your CV where it honestly reflects your experience. If the role asks for AWS experience and you have it, say AWS. Do not assume the system will understand that your cloud infrastructure work was on Amazon Web Services.
Second, use a clean, single-column CV format. Avoid tables, graphics, and multi-column layouts. Use standard section headings like Experience, Skills, and Education. Make sure your CV parses correctly as plain text, not just looks good as a PDF.
Third, be specific about your depth of experience. Instead of listing Python as a skill, describe what you built with it and for how long. This helps both ATS systems that are evolving toward contextual analysis and the human reviewers who will eventually read your CV.
And fourth, get your CV analysed before you apply. Our free CV analysis tool scores your CV against ATS compatibility criteria and highlights specific issues. It takes two minutes and can make the difference between your application being seen or silently discarded.
The Direction the Industry Is Moving
Keyword matching is not going to disappear overnight. But the recruitment technology industry is clearly moving toward skills-based, context-aware matching. Vector search, semantic embeddings, and AI-powered analysis are replacing word counting with meaning comprehension.
The employers who adopt these tools first will see better candidates, shorter hiring cycles, and stronger teams. The candidates who understand how these systems work will present themselves more effectively. And eventually, the technology will be good enough that a qualified candidate with an honest CV should never be silently rejected because they used the wrong synonym.
We are not there yet. But we are closer than most people think.