How ATSs Are Ghosting Great Candidates (And What to Do About It)

Traditional ATS platforms screen candidates by matching resume keywords against job descriptions. The logic is straightforward: the more keywords a resume matches, the more relevant the candidate.

The result is predictable. Candidates have learned to stuff their resumes with keywords that reflect the job description regardless of whether those keywords reflect their actual skills or experience. The practice is widespread enough that LinkedIn News has covered it as a standard job-seeking strategy.

The outcome for recruiters is a diluted applicant pool. The candidates who advance are not necessarily the best fit; they are the most ATS-literate. A separate but related data point illustrates the scale of the problem: according to industry research, up to 75% of resumes never reach a human recruiter at all. For high-volume, low-complexity roles this may be a manageable trade-off. For organizations competing for top talent, it is a significant and largely invisible problem.

A resume is a compressed version of a professional story. Traditional ATS platforms read it as a list of data points. That gap matters more than most hiring teams realise.

Candidates with non-linear career paths, employment gaps, or cross-industry experience routinely fall through ATS filters, not because they lack ability, but because their background does not map cleanly onto a keyword set. As recruiting commentator Andrew Jenkins has noted, traditional ATSs function like a blunt instrument, treating every candidate the same regardless of the nuance in their experience.

This lack of context also drives poor candidate experience. Applicants who receive no acknowledgment, no feedback, and no human contact feel reduced to a data point. That perception spreads. Over time, it erodes employer brand in ways that are difficult and expensive to reverse.

The scale of this problem is well documented. A 2021 Harvard Business School study, referenced consistently in updated research through 2025, found that 88% of employers believe they have lost highly qualified candidates because those candidates were screened out by an ATS. For middle-skilled roles, that figure rises to 94%.

Formatting is a contributing factor that often goes unacknowledged. Traditional ATS platforms struggle to parse resumes that include graphics, tables, or non-standard fonts. A candidate with a visually formatted resume or an unconventional layout can be eliminated from consideration before a human recruiter ever sees their name. The system discards the candidate, not because they are unqualified, but because the document was not optimised for machine ingestion.

The result is a talent gap that organizations create for themselves.

ATS platforms are only as objective as the data used to train them. When historical hiring data reflects existing biases, those biases get encoded into the screening logic. Candidates with certain educational backgrounds, career trajectories, or demographic characteristics may be systematically advantaged or disadvantaged without any deliberate intent.

The consequences can be significant. The Mobley v. Workday case, which a federal judge allowed to proceed in July 2024, became the first major class action lawsuit against AI-driven hiring discrimination in the United States. It illustrates a broader risk that many organizations have not fully examined in their own ATS configurations. A November 2025 study from the University of Washington, presented at the AAAI/ACM Conference on AI, Ethics, and Society, found that when AI systems showed racial bias in hiring recommendations, human reviewers tended to mirror those biases, selecting candidates in line with the AI's preferences. Without AI input, the same reviewers made unbiased choices. The problem, in other words, does not stay inside the algorithm.