Why your job descriptions deter the best candidates.

The job description is where most hiring biases enter the funnel. Not in the screening, not in the interview — the JD. By the time a biased JD has been published, the candidate population that applies is already filtered, and no downstream "blind screening" or "diversity sourcing" can fix it.

This is one of the most well-documented effects in the entire HR literature. It is also one of the least addressed in HR tooling.

The four bias categories that show up in JDs

1. Gendered language. Words like "aggressive," "competitive," "dominant" are statistically more likely to deter female applicants. Words like "collaborative," "supportive," "nurturing" do the opposite. Most JDs unconsciously cluster toward one end.
2. Age-coded requirements. "Digital native," "5+ years using GitHub," "high energy" all carry strong age signal. "Recent graduate" excludes returners. "Senior" is sometimes a real seniority signal, sometimes a code for "expensive."
3. Unrealistic combinations. "10 years experience in Kubernetes" (which is 11 years old). "Expert in React and Vue and Angular and Svelte" (which excludes anyone who chose to specialise). "PhD plus 5 years industry plus startup experience" (a Venn diagram with very few people in it).
4. Implicit demographic signals. Location requirements that exclude visa categories, education requirements that exclude returners, language requirements that exclude non-native speakers when fluency isn’t actually needed.

The data

Studies from LinkedIn, Textio, and academic HR research consistently show:

• JDs with gendered language reduce female applicants by 5–15%
• JDs with age-coded language reduce applicants over 40 by 20–30%
• JDs with unrealistic combination requirements reduce all applicants by 40–60% — not because the requirements are real, but because qualified people self-screen out

The single biggest lever for improving the diversity of your candidate pipeline isn’t in the sourcing stage. It’s in the wording of the JD itself.

What the bias auditor does

Before any JD is published in CognitoHire, the bias auditor runs. It checks for:

• Gendered word patterns (using validated lexicons from academic HR research)
• Age-coded requirements (with explicit suggestions for neutral alternatives)
• Requirement combinations that statistically exclude qualified candidates
• Education requirements that aren’t justified by the role’s actual skill needs
• Location and visa language that may be unnecessarily exclusionary

The auditor doesn’t block publication. It surfaces concerns to the recruiter, with citations to the underlying research. The recruiter chooses. The reasoning is logged. The board has an audit trail. The legal team has a defensible record.

A small example

Real JD (anonymised) submitted by a customer last quarter:

Senior Backend Engineer. 8+ years experience. Aggressive self-starter. Recent CS graduate from a top-tier programme. Comfortable working long hours in a high-energy startup environment.

The bias auditor flagged: combination unrealistic (8+ years AND recent graduate), age-coded ("high energy," "long hours"), gender-coded ("aggressive"), education-narrow ("top-tier programme"). Estimated reduction in qualified applicant pool: 60–75%.

The customer rewrote. The applicant pool tripled. The eventual hire was a 14-year-experience engineer from a state university who would never have applied to the original posting.