JD Optimizer: Aligning Expectations with Market Reality

The Problem No One Talks About at the Top of the Funnel

Most hiring failures are diagnosed at the offer stage or during onboarding. The more common origin is earlier — in the job description itself. A posting that describes a unicorn candidate, stacks credential requirements beyond what the role actually demands, or uses exclusionary language will quietly distort every downstream step: sourcing, screening, interviewing, and ultimately the quality of the hire.

This is not a minor inefficiency. Research on job requirements inflation is consistent: Burning Glass Technologies (now Lightcast) documented in their 2014 analysis Moving the Goalposts that employers routinely require a four-year degree for roles where fewer than 20% of current incumbents hold one. That misalignment drives unnecessary credential gatekeeping, narrows candidate pools without improving quality, and — as subsequent SHRM research has noted — extends time-to-fill without a corresponding improvement in retention or performance.

The question worth asking is: what would it look like to treat the job description as a calibration instrument rather than a wish list?

A Realistic Scenario: The Over-Specified Operations Role

Consider a mid-sized logistics company posting a Supply Chain Analyst role. The hiring manager, drawing on memory of a high-performing predecessor, builds a description that includes:

• 7+ years of experience in supply chain analytics
• An MBA or Master's in Operations Research
• Proficiency in SAP, Tableau, Python, and SQL
• Familiarity with Six Sigma and APICS CSCP certification preferred
• Demonstrated experience managing cross-functional teams

The actual role: an individual contributor analyst position expected to build dashboards, maintain inventory models, and flag anomalies for a senior team to act on. Team management is not in scope for at least 18 months.

The description as written will screen out qualified mid-career analysts who lack the MBA but have the hands-on technical skills. It will attract over-qualified candidates who will leave when the scope becomes clear. And it will generate a legally unnecessary credential filter that, depending on how the pool breaks down demographically, may invite disparate impact scrutiny under Griggs v. Duke Power Co. (1971) — a Supreme Court precedent establishing that facially neutral requirements with discriminatory effect require demonstrable job-relatedness.

This is the problem an AI job posting improver, used correctly, is designed to surface.

What a Structured JD Optimization Framework Actually Does

Optimization here does not mean softening requirements to attract more applicants. It means aligning stated requirements with three evidence-anchored benchmarks:

1. Role-Task Congruence

Every listed requirement should map to a specific task or outcome in the role. If a requirement cannot be tied to at least one job function, it is a proxy — and proxies introduce both legal risk and screening inefficiency.

The ONET database (maintained by the U.S. Department of Labor) provides validated task taxonomies for hundreds of occupational categories. A structured optimizer cross-references stated requirements against ONET task clusters for the relevant occupation. Requirements that appear in fewer than 30% of similar role definitions in the database warrant a flag.

2. Market Calibration

Labor market data determines whether the stated experience bands are realistic for the candidate population available. Lightcast (formerly Burning Glass) and LinkedIn Talent Insights both publish real-time data on candidate supply at various experience and credential levels. An optimizer that ingests this data can tell you: at the salary band you have budgeted, the candidate profile you have described represents approximately 3% of available candidates in your metro area. That is actionable information.

Lack of market calibration is a primary driver of the "purple squirrel" phenomenon — a position that sits open for months because the specification, not the market, is broken.

3. Language Audit for Accessibility and Bias

The third function of a JD optimizer is linguistic. Word choice in job postings correlates with applicant pool demographics. Gaucher, Friesen, and Kay (2011), published in the Journal of Personality and Social Psychology, showed experimentally that job postings using masculine-coded language ("dominant," "competitive," "ninja") attracted fewer female applicants, independent of the actual job content. This effect held even when respondents could not consciously identify the bias in the language.

A structured language audit flags:

• Gendered adjectives and verbs
• Unnecessarily exclusionary credential language ("top-tier university")
• Vague superlatives that communicate aspiration rather than function ("rockstar," "exceptional")
• Readability scores that exceed the literacy level needed for the role (Flesch-Kincaid grade level is a usable proxy)

How Verdict Implements This Framework

Verdict approaches the job description as the root node of the evaluation graph. Before candidates are scored, the role specification itself is interrogated against the three benchmarks above.

When a hiring team submits a JD, the optimizer surfaces a calibrated requirement set — distinguishing between threshold requirements (must-have for minimum job performance), differentiating requirements (predict performance above threshold), and aspirational requirements (nice-to-have but not predictive). This taxonomy is grounded in the validity evidence hierarchy established by Schmidt and Hunter (1998) in Psychological Bulletin, which remains the most comprehensive meta-analytic summary of what predicts job performance.

The output is not a rewritten job description handed back as a fait accompli. It is an annotated markup showing:

• Which requirements have strong job-relatedness evidence
• Which are market-misaligned at the posted compensation
• Which carry potential disparate impact exposure under validated bias research
• What the estimated candidate supply looks like after each adjustment

The hiring manager retains decision authority. The optimizer provides evidence. That distinction matters.

Returning to the Scenario: What Changes

Back to the Supply Chain Analyst posting. After running the role through the framework:

The revised description will reach a materially larger qualified candidate pool, reduce time-to-fill, and produce a screening shortlist that is more defensible if the decision is ever audited. It will also be a more honest document — and honest job descriptions attract candidates who take the job for what it actually is, which is a retention predictor in its own right.

What This Framework Does Not Do

It is worth being explicit about the limits of JD optimization:

• It does not guarantee diverse applicant pools. Language and credential calibration reduce unnecessary barriers; they do not substitute for sourcing strategy. A well-calibrated posting distributed through a narrow channel still produces a narrow pool.
• It does not replace domain judgment. The optimizer flags misalignment; the hiring manager owns the decision about whether a given requirement is truly threshold or merely familiar.
• It is not a proxy for structured interviewing. An optimized JD improves top-of-funnel inputs. Evaluation quality at the interview stage is a separate, though related, problem. See Forensic Interviewing: Structured Kit Generation for how structured interview design connects to the requirement taxonomy established here.

For teams that want to understand how the requirement set maps to actual candidate scoring, Candidate Evaluation Criteria: How to Score Candidates provides a complementary methodology for converting JD requirements into a weighted evaluation rubric.

The Downstream Case for Getting This Right

The evidence on over-specification costs is not soft. A Harvard Business School and Accenture joint report (Hidden Workers: Untapped Talent, Fuller et al., 2021) surveyed over 8,000 employers and found that automated screening systems — calibrated to inflated JD requirements — were systematically filtering out 27 million workers who were qualified to perform the roles in question. The cost is borne by employers in prolonged vacancies and by workers who are screened out before a human ever sees their application.

An AI job posting improver, embedded early in the hiring workflow, addresses this at the source. The effort required is a single structured review before the posting goes live — a leverage point that compounds across every application that follows.

See It in Practice

If you want to see how Verdict evaluates candidates against a calibrated requirement set — rather than an inflated wish list — you can run a side-by-side candidate comparison using your own role and materials. No setup required, no pressure. Just the methodology applied to your actual hiring context.