Blog
Latest
Your diversity dashboard shows a problem:
Application stage: 35% candidates from underrepresented groups Shortlist stage: 18% candidates from underrepresented groups Hire stage: 12% hires from underrepresented groups
You increased diversity sourcing. You partnered with minority-serving institutions. You sponsored recruitment events at HBCUs. Your application volume from diverse candidates is strong.
So why is diversity disappearing between application and shortlist?
The answer is hiding in plain sight: keyword matching.
Your screening process—designed to be "objective" and "bias-free"—is systematically excluding diverse candidates. Not because they lack capability. Because they describe their experience differently.
How Keyword Screening Creates Systemic Exclusion
Traditional resume screening filters for exact matches:
Specific job titles ("Product Manager II" not "Product Lead")
Buzzword phrases ("Agile methodology" not "iterative development")
Pedigreed backgrounds ("target schools" not state universities)
Linear career paths (no gaps, no pivots, no non-traditional routes)
These criteria feel neutral. They're not. They're exclusionary by design.
Here's why:
1. Keyword matching favors pattern recognition over potential
Candidates from non-traditional backgrounds often describe equivalent experience using different language:
A career-switcher from teaching might say "curriculum design" instead of "learning program development"
A first-generation college graduate might list "community organizer" instead of "stakeholder management"
A veteran might say "mission planning" instead of "strategic project management"
The capabilities are identical. The keywords aren't. And keyword-based ATS systems can't tell the difference.
Research from Harvard Business School's "Hidden Workers" study found that 88% of employers report losing qualified candidates due to screening configuration issues. The study specifically highlighted that screening systems exclude candidates who have the skills but describe them differently than job descriptions expect.
2. Pedigree filters systematically exclude non-traditional talent
Many ATS configurations prioritize:
Graduates from "target schools" (which disproportionately serve affluent, white students)
Previous employment at "name brand" companies (which have historically had homogeneous workforces)
Specific degree programs (which may be financially inaccessible to first-generation students)
A candidate who attended a state school, worked at smaller companies, and built skills through non-traditional paths might be more qualified than a candidate with the "perfect" pedigree. But keyword screening filters them out before a human ever sees their application.
Organizations using predictive analytics achieve 39% fairer hiring treatment for women and 45% fairer treatment for racial minorities compared to traditional keyword approaches. The difference? They screen for demonstrated capabilities, not pedigree proxies.
3. Employment gaps trigger automatic exclusion
Many ATS systems are configured to deprioritize or auto-reject candidates with employment gaps longer than six months.
Who has employment gaps?
Parents (disproportionately women) who took time for caregiving
People who faced layoffs during economic downturns (disproportionately affecting Black and Latino workers)
Immigrants navigating credential recognition in a new country
People with disabilities managing health challenges
Justice-impacted individuals reentering the workforce
These gaps don't predict poor performance. But keyword screening treats them as disqualifying.
Recent research on ATS systems reveals that while only 8% of recruiters configure content-based auto-rejection, 92% use filters and scoring that effectively deprioritize non-traditional candidates. The result is the same: systematic exclusion dressed up as "objective" screening.
The Distance Traveled Blind Spot
Medical schools faced this exact problem two decades ago. Their admissions criteria—GPA, MCAT scores, undergraduate prestige—were "objective." They were also exclusionary.
Students from disadvantaged backgrounds often had slightly lower test scores not because they lacked capability, but because they:
Worked full-time while attending school
Attended under-resourced high schools
Were first in their families to attend college
Faced systemic barriers that privileged students didn't encounter
Traditional screening missed these candidates entirely.
The breakthrough came when medical schools introduced "distance traveled" evaluation: How far has this candidate progressed relative to their starting point? What obstacles did they overcome? What capabilities did they develop along the way?
Students admitted under distance traveled frameworks often matched or outperformed their higher-credentialed peers on metrics that actually mattered: clinical judgment, resilience under pressure, patient outcomes.
Why? Because overcoming obstacles builds exactly the skills that predict success in complex, ambiguous environments: learning agility, creative problem-solving, persistence through setbacks.
Your screening process misses this entirely—because keyword matching can't measure distance traveled.
What This Costs You
The diversity impact is measurable:
Research shows that diverse teams:
Make better decisions 87% of the time (Cloverpop study)
Are 35% more likely to outperform homogeneous competitors (McKinsey)
Generate 19% higher revenue from innovation (BCG research)
When keyword screening filters out diverse candidates, you're not just missing diversity goals. You're actively excluding the perspectives and problem-solving approaches that drive innovation and performance.
For a mid-market company, this translates to millions in lost innovation value and competitive disadvantage.
And it's completely avoidable.
How to Fix Screening for Diversity
Organizations that have solved diversity screening follow three practices:
1. They screen for transferable skills, not exact keywords
Instead of filtering for "5 years as Product Manager," they evaluate:
Strategic thinking ability (evidenced through work samples, case studies)
Learning agility (demonstrated by pivots, skill acquisition, adaptation)
Problem-solving under ambiguity (shown through project examples, portfolio work)
This opens the aperture to candidates who built these skills through non-traditional paths.
2. They measure distance traveled, not just arrival point
They ask: What challenges did this candidate overcome to get here? What did they build from limited resources? How did they navigate obstacles?
A candidate who taught themselves to code while working full-time and supporting family demonstrates learning agility, resourcefulness, and persistence—often better predictors of success than a candidate who took a traditional path through privileged circumstances.
3. They audit screening for demographic pass-through rates
They track: At what rate do candidates from different backgrounds advance through screening?
If women, people of color, or candidates from non-traditional backgrounds pass through at significantly lower rates when controlling for qualifications, the screening criteria are creating systemic exclusion.
This data makes the invisible visible—and actionable.
Start With Diagnosis
If your diversity falls off significantly between application and shortlist, your screening process is the problem.
Download our free Screening Quality Audit to identify where keyword matching and pedigree filters are systematically excluding qualified diverse candidates.
The 15-point diagnostic helps you:
Identify screening criteria that create systemic exclusion
Measure demographic pass-through rates at each screening stage
Build screening methods that evaluate transferable skills and distance traveled
Audit whether your ATS configuration deprioritizes non-traditional candidates
DOWNLOAD THE SCREENING QUALITY AUDIT
Your diversity sourcing is working. Your screening is the bottleneck.
Keyword matching isn't objective. It's exclusionary by design.
Fix your screening criteria, and you'll fix your diversity pipeline.