Problems With Traditional Resume Screening Systems

After 35 years in positions where I've been a hiring manager or part of hiring teams, I need to say something that's been weighing on me - We've lost the plot in recruitment. Throughout my career, I've always operated on one principle, I hire people, not CVs. But somewhere along the way, we've let AI screening and automated systems become the gatekeepers. We've traded human judgment for algorithms. We've replaced phone conversations and face-to-face meetings with keyword matching and automated rejections. And we're paying a price for it. Here's what these systems can't capture: • The transferable skills hiding in an unconventional background • The drive and potential in someone making a career pivot • The genuine reasons why someone wants to move, escaping a toxic environment, seeking stability, looking for a place where they're valued and trusted, not micromanaged • The human instinct that tells you "this person has something" even if their CV isn't textbook perfect Every person deserves a chance. Not just the ones whose resumes happen to tick all the right boxes in an ATS. Some of the best hires I've made over three decades were people who didn't look perfect on paper. They looked perfect in conversation. They had the hunger, the attitude, the capability, things you only discover when you actually talk to someone. We're missing exceptional talent because we've made efficiency and scale more important than insight and judgment. So here's my challenge to fellow hiring managers: Push back where you can. Insist on human screening. Set your filters wider. Create ways for real people to have real conversations before automated systems shut the door. Because at the end of the day, businesses are built by people, not CVs. And the best talent isn't always found in perfect formatting, it's found in genuine conversation. Who else thinks it's time we brought humanity back into hiring? #Recruitment #HiringManagers #Talent #Leadership #HumanResources #PeopleFirst

A recent study from micro1’s research lab, "Uncovering Candidate Potential with Multi-Modal AI Assessments", confirms what many have long suspected: resumes are a noisy and inefficient tool for talent allocation. When Nima Yazdani and Rumi Allbert compared traditional resume-based evaluations with an AI-driven “Match Score” which factors in interviews, problem-solving, and role-specific assessments, the AI approach proved both more predictive and consistent. The numbers speak for themselves: the average AI score was higher (66.9 vs. 59.7), displayed greater stability, and showed a much tighter range than resume-based scores. Put simply, resumes are not only poorer at predicting hiring outcomes, but they also inject considerably more variability into the selection process. Why does this matter for the broader economy? Every mismatch imposes real costs. “Hidden gems”: candidates who may not stand out on paper but excel in AI assessments are too often overlooked, resulting in lost potential for productivity and innovation. Conversely, “inflated resumes” that mask weaker capabilities cost companies time and resources interviewing unqualified candidates. With a weak correlation between the two approaches (r = 0.19), relying solely on resumes amounts to misallocating human capital, the most valuable resource in the modern economy. The takeaways? A couple... 1) Multi-modal AI assessments aren’t just a fairer approach; they make labor markets more efficient. By reducing noise in hiring and improving the accuracy of matches, companies can tap into overlooked talent and avoid costly hiring errors. That’s not just a hiring upgrade, but an economic one. 2) If AI assessments can identify strong candidates more effectively than resumes, maybe it’s time to question whether the traditional CV should still be at the center of hiring decisions. Link to summary and downloadable paper here: https://lnkd.in/dRQm7ppv

I reviewed 547 resumes for a Data Scientist role on my team and found that 53 had parsing issues with the Applicant Tracking System (ATS). Here’s what I observed and some practical solutions: Problem 1: ATS Struggles with 'Modern' Resume Styles 9 resumes were blank or had no content. I believe this is due to: 🔸 These resumes often used fancy Word templates with two-column layouts. For example, one column had contact details, education, and skills, while the other had work experience.  I believe we can avoid this by 🔹 using a simple, clean formatting (LaTeX works well for this) 🔹 submitting resumes in PDF format rather than Word documents 🔹 sticking to a one-column layout to ensure clarity 🔹 maintaining consistent margins (minimum 0.5 inches) and line spacing (1.0 minimum) Problem 2: Distorted Content 44 resumes were harder to read. I believe this is due to: 🔸 Use of icons, images, charts, tables, or colors 🔸 "Justify" text alignment, which creates inconsistent word spacing 🔸 Fonts with ligatures (e.g., "fi", "fl", “ft”, etc), where letters merge and confuse the ATS (e.g., "artificial" becoming "arti cial") 🔸 Special characters like apostrophes (') and ampersands (&) may not render correctly I believe we can avoid this by  🔹 using fonts like Arial or Calibri and avoid italics 🔹 disabling ligatures: In Word: Select the text, go to Text Effects -> Ligatures -> None In LaTeX Option 1: Use the microtype package to disable ‘f-ligatures’: \usepackage{microtype}  \DisableLigatures[f]{encoding = *, family = *} Option 2: Globally disable ligatures with: \input{glyphtounicode}  \pdfgentounicode=1 🔹 using “Left Align” instead of “Justify” for text alignment 🔹 spelling out “and” instead of using the ampersand (&) Note: In my experience, ATS systems are tools for coordinating applications and do not auto-reject resumes based on formatting. Auto-rejections occur when specific rules are set by the hiring team, such as answering “no” to a key question like, “Do you have 5+ years of experience with R?” Be cautious of services that sell “ATS-compatible” resume templates or promise better ATS scores. These services may intentionally lower your scores to sell their premium services. It’s best to focus on clear, simple formatting and content that aligns with the job description. #resume #ats

THE SCREENING PROBLEM Human CV Screening vs Algorithmic Screening Yesterday I showed that the famous "7-second CV scan" is a debunked marketing stat, and that what recruiters actually do during screening is structural pattern matching against an unvalidated mental template. Today: what happens when we automate that? There are broadly three classes of AI screening systems in use right now. Deterministic systems apply hard rules: must have a degree X and Y years of experience. They are transparent, and they do exactly what you tell them, which is the problem, because what you tell them is usually a recycled job spec that nobody pressure-tested. Statistical and machine learning models learn from historical hiring data. Raghavan et al. (FAT* Conference, 2020) showed these systems reliably replicate existing biases in that data, including biases the organisation did not know it had. LLM-based evaluators are the newest class. Qin et al. (2023) found that they can hallucinate qualifications, infer traits from formatting, and produce confident assessments that lack grounding in what the candidate actually wrote. All three share the same structural vulnerability. They are only as good as the specifications they are given. And in most organisations, that specification is the same brief that was producing template-driven human screening in the first place. But here is the part that should worry TA leaders most. AI does not just replicate the template. It drifts. When a model recommends candidates, recruiters accept them, and those candidates enter the training data, the model retrains on a narrowing sample. Over successive cycles, the effective definition of a "good candidate" contracts without anyone changing the specification. Sculley et al. (NIPS, 2015) documented this pattern as feedback loop bias in machine learning systems. The recruiter sees only current recommendations. The TA leader sees only throughput dashboards. Nobody monitors whether the model's working definition has silently shifted. Now multiply that across a market. Jobscan reports 97.8% of Fortune 500 companies use an ATS. When multiple employers use the same vendor's AI, drift propagates across the entire client base simultaneously. One vendor's model narrows, and an entire sector's talent pipeline narrows with it. A candidate excluded by one employer's AI is excluded across the market, not because independent evaluators agreed, but because the same model reached the same conclusion everywhere. The EU AI Act (2024) classifies employment AI as high-risk. But it does not yet address cross-organisational model propagation. Nobody is talking about this. They should be. Thursday: what candidates are doing about it, and why it makes everything worse.

A recruiter showed me what a Canva resume looks like in their ATS. It parsed as: "JOHN EXPERIENCE SKILLS SMITH MARKETING 2021 MANAGER EDUCATION ADOBE PRESENT." That candidate spent hours perfecting the design. The recruiter spent 3 seconds moving to the next application. The problem isn't that people are trying to stand out - it's that most job seekers don't know how ATS actually works. Modern systems use Natural Language Processing. They scramble two-column layouts. They can't parse text boxes or graphics. They flag resumes with "Marketing Manager" repeated 15 times as spam. Yet every day, talented professionals are using beautiful templates and keyword stuffing strategies that backfire instantly. I tested hundreds of resumes through actual ATS systems. Resumes with the exact job title from the posting were 10.6 times more likely to get interviews. Simple single-column formats beat every creative template. Natural keyword placement throughout experience sections outperformed desperate stuffing every time. Many talented professionals get auto-rejected for roles they're perfect for. Not because they lack qualifications, but because they think they need to game a system instead of clearly showing their value. I built a free ATS checker to show people exactly what recruiters see. No templates to sell. Just transparency about whether your resume actually works. Comment ATS and I will share it with you. What "professional" resume template destroyed your job search?

I've been hearing increasing frustration from my network recently and my inbox is filled with outstanding job seekers looking for advice, about perfectly qualified candidates, especially seasoned leaders, being auto-disqualified or ghosted almost immediately at application stage, often within minutes or hours. Many point the finger at AI-powered screening tools as the main culprit in this tough job market. While that's possibly part of the picture, my sense is that the issue runs deeper than just "the bots are broken." I'm particularly struck by stories from highly experienced executives and senior leaders whose resumes align almost word-for-word with the job description, yet upon application, they hit an immediate wall: ATS rejection, "disqualified" status, or complete silence. In a market where internal recruiters and hiring teams are drowning in applications (often hundreds per role), how and why are these standout candidates, who most organisations would jump at the chance to interview, falling at the very first hurdle? Common triggers seem to include: Knockout/pre-screening questions in the application form (e.g., exact years of experience, specific certifications, salary expectations) a single "no" can auto-reject before the resume is even parsed. Subtle mismatches in keyword phrasing (even when the experience is equivalent and cvs have been written to ensure they're ATS friendly). Hidden filters like minimum requirements, or even over-qualification in some rigid systems. High-volume overload leading to stricter automated cutoffs. What are you seeing on the ground? Is it truly AI gone wrong, knockout questions biting back, or something else entirely in this volatile hiring environment? Would love to hear from recruiters, hiring managers, and job seekers, what's really happening here, and how are top talent candidates navigating it successfully? #Hiring #Recruitment #JobSearch #Leadership #ATS #TalentAcquisition

UNCOMFORTABLE TRUTH #7: The best candidate probably isn’t in your finalist pool. This one keeps me up at night. Right now, somewhere out there, there’s a leader who could transform your school — who has the mindset your students need, the trust your staff needs, the clarity your community needs. And you never met them. Why? Because we’ve inverted the process. We filter first, and we filter with tools that don’t measure what actually predicts success. Those high-potential candidates — the ones who think differently, who lead courageously, who could move your system forward — are often screened out in the first round. ❌ Wrong alma mater. ❌ Too few “acceptable” years of experience. ❌ Nontraditional career path. ❌ Didn’t use the right buzzwords in their cover letter. Here’s what the research shows: Resume screening eliminates 70–80% of potentially strong candidates based on surface-level criteria that have minimal correlation with actual performance. That “rigorous screening process” isn’t ensuring you advance the strongest candidates. It’s ensuring you advance the candidates who are best at navigating the filtering system. Let’s be honest: The finalist pool often represents— • People who checked every credential box. • People who wrote exactly what you wanted to read. • People who look like the leaders you’ve hired before. It does not represent— • The full spectrum of capable leaders. • Candidates with unconventional paths to excellence. • Voices and perspectives that could challenge your status quo. And the person your school actually needs? They’re often the ones who never made it past round one. Not because they lacked talent—but because they didn’t fit the filters. This is the most expensive filter you have. And it’s being applied first. So what can we do? ✅ Screen for minimum qualifications, not maximum credentials. ✅ Cast a wider net and evaluate more candidates through demonstrated performance, not just paper filters. ✅ Shift your investment: less time reviewing resumes, more time watching candidates think, decide, and act. This concludes our “7 Uncomfortable Truths About Education Hiring” series. If these truths made you nod, pause, or squirm a little — that’s the point. We can’t fix what we refuse to name. TruFit Talent was built to confront these truths head-on — transforming how schools identify, grow, and retain talent. We replace intuition with evidence, bias with equity, and chance with confidence. Because the future of education depends on how we choose its people. #TalentStrategy #EducationLeadership #HiringInnovation #FutureOfWork #SchoolLeadership #TruFitTalent #RecruitDevelopRetain #TruFitTalent

What resumes actually select for is not competence, but legibility. The modern resume endures not because it is accurate, but because it is familiar. It reduces a working life to a page and invites us to treat coherence as capability. Over time, this habit has become institutionalized, rarely questioned, and widely defended, despite mounting evidence that it performs poorly as a measure of future contribution. Economic theory has long warned that when performance is difficult to observe, selection systems favor those who can signal rather than those who can execute. Hiring is a textbook case. Empirical research in industrial and organizational psychology consistently shows that credentials and self reported experience have limited predictive value, while work-sample evaluations and task based assessments produce materially stronger correlations with real performance. Yet resumes remain the primary gatekeeper. Even large, analytically sophisticated firms have acknowledged this limitation. Internal analyses shared by companies such as Google, IBM, and Accenture have shown that traditional markers - pedigree, prior titles, even years of experience - decay quickly as predictors once someone enters the workforce. What matters thereafter is not where someone has been, but how they perform under constraint. The persistence of the resume is therefore not an evidentiary choice, but an operational one. It is efficient to screen, easy to standardize, and defensible after the fact. In exchange, it quietly privileges narrative fluency, institutional familiarity, and social confidence. The system selects for those who are legible to it. The cost of this misalignment is subtle but cumulative. Quiet operators are overlooked. Late bloomers stall. Teams absorb variance that could have been avoided. And organizations compensate by adding more interviews, more panels, more subjective checkpoints, mistaking volume of judgment for quality of signal. In nearly every other high stakes domain, we have abandoned this approach. Engineers are evaluated by systems that run. Investors by realized returns. Researchers by results that replicate. Hiring remains an outlier, still grounded in inference rather than observation. I explore this argument, and the historical reasons it persists, here: 👉 https://lnkd.in/eQQfD7F2 This line of inquiry is also what led us to build Velric. We are launching our public beta on December 21, introducing personalized, real world work simulations that evaluate candidates on demonstrated execution rather than narrated experience. The beta is free and available worldwide. The goal is not to discard history, but to stop mistaking it for evidence. If hiring is a decision about future performance, then it deserves systems designed to observe performance directly.

I recently spoke with a dedicated job seeker who applied to 377 roles on LinkedIn. She was strategic — targeting only fresh postings and tailoring her resume every time. She did everything “right.” Yet astonishingly, only 27 of those applications were even viewed by recruiters; 350 got no response at all (no LinkedIn alerts, no emails, nothing). 377 applications submitted (all carefully customized). Only 27 were viewed; 350 were ignored. 2 interviews resulted (one with an AI chatbot, one panel interview); 0 job offers. This is not a candidate problem — it’s a hiring process problem. The System Overwhelmed Automated tools have created an avalanche of applications. One report found application volume surged by 48% year-over-year as candidates use “easy apply” bots. Hiring teams are swamped: it’s “fundamentally broken when your ATS is flooded with 500 applications for a single role and only 20 of them are remotely qualified” No surprise then that 66% of job seekers report feeling burned out by the search Many qualified people give up or question themselves — not because of their skill level, but because the system is impersonal and slow. For too long, recruiters have measured success by the wrong metrics. As one expert puts it, “time-to-fill, cost-per-click, and raw application numbers tell us very little about whether we’re actually hiring the right people.” Over-automated: Candidates are screened by algorithms with no human interaction. Under-human: Automated interviews and ghosting offer no empathy or feedback. Slow & Broken: People are left hanging indefinitely or completely ignored. People Are Not Pipelines Remember: job seekers are human beings, not funnel metrics. Treating applicants like numbers dehumanizes the process. If great candidates are burning out, it’s time to fix your process, not blame them. Respect: A timely personal response (even a rejection) shows you value candidates. Clarity: Simplify the application (35% of candidates abandon long applications). Humanity: Balance tech with real connection — many candidates trust humans more than bots 📌 FREE Resume Review: To help job seekers navigate this broken system, I’m offering a completely free resume review. Follow or connect with me and send your resume — I’ll give you personal feedback and support. You deserve better treatment in your job search. #JobSeekers #Recruitment #Empathy #Hiring