Knowitol Research — Original analysis
What ATS systems actually reject — analysis of resumes scanned on Knowitol
An anonymized look at how often resumes fall short of the 75-point match threshold, and the gap categories driving rejection.
Executive summary
Most resumes uploaded to Knowitol score below the threshold ATS-aware recruiters use as a soft cutoff, and the reason is rarely formatting — it is missing role-specific signals. Across hundreds of thousands of completed scans, the most common gap categories are cloud/devops tooling, programming frameworks, and data/AI vocabulary. Generic resume tips that focus on templates miss this entirely.
Methodology
We pulled every completed resume scan in the last 12 months from our production database, excluded internal test traffic (is_fake users), and computed score distributions and gap categorizations directly from each scan's analysisResult. Gap categorization uses a deterministic keyword classifier that buckets each missing keyword into one of ten skill categories. Cells with fewer than 50 scans are suppressed.
Score distribution: where most resumes actually land
The biggest category of resumes scanned on Knowitol does not score 90+ — it scores in the 60–74 'borderline' band. That is the band where ATS-aware recruiters report the highest variance in human-screen outcomes: a borderline-scored resume might pass an automated keyword filter but still fail a 6-second human glance, or vice versa.
What is actually missing from rejected resumes
When we categorize the missing keywords detected by Knowitol's scanner, the largest gap categories are not soft skills or formatting — they are concrete, role-specific technical signals. Cloud and DevOps tooling, programming frameworks, and modern data/AI vocabulary dominate. This is the strongest evidence that 'ATS-friendly templates' alone do not solve the rejection problem.
No data for chart "Top reasons resumes fall short — gap categories".
Why this matters
- A resume sitting at 60–74 is not 'ATS-broken'. It is missing role-specific signals the recruiter is using to triage.
- The gap categories we see at scale are stable across quarters — meaning generic 'add more keywords' advice is not enough; the keywords have to map to the actual role being targeted.
- Tools that only check formatting (margin width, font, parsability) cannot detect or close the gaps that drive most below-threshold scores in our data.
The biggest myth in resume advice is that ATS rejection is about formatting. In our data it is overwhelmingly about role-specific signal density.
Sources
- Knowitol production scan database (2026)
- Knowitol gap-category classifier (deterministic, open methodology) (2026)
Frequently asked questions
How do you make sure individual users are not identifiable?
Every published statistic is computed across at least 50 records (our minimum cell size). Cells with fewer than 50 records are suppressed entirely. We never publish per-user, per-resume, or per-employer data — only category-level aggregates. Internal test traffic (is_fake users) is excluded from every query.
Can journalists cite this study?
Yes. All Knowitol Research data is released under CC BY 4.0 — free to cite, quote, or reproduce with attribution to Knowitol Research and a link to the source report. The press kit at /research/press includes a copy-pasteable boilerplate.
How often is this report updated?
Aggregations are re-computed on every publish from the live scan database, and we re-publish quarterly. Each report shows the exact generation timestamp under Methodology.
Cite this report
Research, K. (2026, April 21). What ATS systems actually reject — analysis of resumes scanned on Knowitol. Knowitol Research. https://knowitol.com/research/ats-rejection-analysis-2026