How Ghost Scores Work

Every Ghost Score is computed from publicly available U.S. government data combined with text analysis of the job posting. Here's exactly what we measure and why.

The Formula

Ghost Score = Text Analysis (60%) + Company Track Record (40%)

60/40 when the employer has significant filing history (>10 PERM filings)

90/10 when the employer has minimal history (≤10 PERM filings)

100/0 when no employer match is found (text analysis only)

Scores are clamped to the 0-100 range. Higher = more likely real. Lower = more likely ghost.

Why we publish this

Most scoring tools hide their methodology behind "proprietary AI." We don't. Every Ghost Score is computed by the formula above, from public U.S. government data, and you can verify it yourself. Proprietary AI scoring is how bad products hide from scrutiny. We'd rather you trust the math than trust our marketing.

Why the 10-filing threshold? Below 10 PERM filings, a company's track record is statistically thin — the sample is too small to draw meaningful conclusions. So we weight the text analysis more heavily and let the job posting speak for itself.

Text Analysis

A forensic analysis of the job posting itself — PERM boilerplate detection, salary vs prevailing wage, duty specificity, technology stack coherence, contract language, and more. See full signal list below.

Company Track Record

The employer's DOL PERM filing history — approval rates, repost patterns, WARN Act notices, expiry rates, and unique role counts — blended into a 0-100 score. See full methodology below.

Our Data Sources

Ghost Scores are built on four federal data sources that anyone can verify independently:

📋

DOL PERM Filings

Permanent labor certification applications filed by employers to sponsor foreign workers. Published quarterly.

374,000+ records dol.gov

⚠️

WARN Act Notices

Federal mass layoff notifications filed by companies planning significant workforce reductions.

Updated daily warnfirehose.com

💰

DOL Prevailing Wages

Occupation-specific wage data by geography. Used to detect salary manipulation in PERM postings.

830+ occupations flcdatacenter.com

📊

DOL LCA / H-1B Filings

Labor Condition Applications filed for H-1B visa workers. Shows employer sponsorship patterns.

600,000+ records dol.gov

All data sources are public domain U.S. government records. We do not scrape LinkedIn, Indeed, or any job board. Users paste job descriptions themselves.

Visa Intelligence Methodology

Our Visa Intelligence features combine DOL PERM data with USCIS H-1B Employer Data Hub records and Visa Bulletin history to give immigration-aware job seekers unprecedented employer insights.

EB Category Classification

We classify each PERM filing as EB-2 or EB-3 using the statutory definitions from INA S203(b) (Immigration and Nationality Act):

EB-2: Master's degree or higher, OR Bachelor's + 5 years progressive experience

EB-3 Professional: Bachelor's degree with less than 5 years experience

EB-3 Skilled Worker: 2+ years training/experience, no degree required

EB-3 Other Worker: Less than 2 years experience

Caveat: Actual I-140 filings may use the EB-3 downgrade strategy when the EB-3 queue moves faster than EB-2. Our classification reflects the PERM application, not the final I-140 category.

Green Card Timeline Formula

T_total = T_PWD + T_Recruit + T_PERM + T_I140 + T_Queue + T_I485

T_PWD: ~6 months (DOL wage determination)

T_Recruit: 3 months (employer recruitment)

T_PERM: 16.5 + audit adj. (DOL processing)

T_I140: 0.5-8 months (USCIS petition)

T_Queue: Variable (visa bulletin backlog)

T_I485: ~7 months (adjustment of status)

Queue estimates use attrition-adjusted projections from Cato Institute backlog methodology and historical Visa Bulletin movement rates from DOS.

Immigration Report Card (A-F Grade)

Each employer receives a letter grade based on 6 weighted components (100 points max):

H-1B Approval Rate (25 pts): 3-year initial petition approval rate from USCIS Data Hub

Wage Premium (20 pts): How much above prevailing wage the employer pays

Retention I:C Ratio (20 pts): Initial vs. Continuing petitions - low ratio = workers stay longer

PERM Expiry Rate (15 pts): Percentage of PERM applications that expire unused

H-1B Independence (10 pts): Whether the employer is classified as H-1B dependent

Wage Level Distribution (10 pts): Percentage filing at Level I (entry) prevailing wage

Verify Our Data

Every data point in Hire4Real.fyi can be independently verified using these public sources:

Two Scores, One Answer

Every scan produces two component scores that blend into your final Ghost Score:

Text Analysis Score

What the job posting says

We analyze the actual words in the posting - specific duties vs vague buzzwords, salary transparency, application method, benefits detail, selling language vs demand-only language, and patterns associated with PERM compliance postings (like requiring paper applications or hyper-specific education requirements). A well-written posting with specific duties, named team, listed salary, and strong benefits scores high. A vague posting full of buzzwords scores low.

Company Track Record

What the employer has done

We look at the employer's actual filing history with the Department of Labor. How many PERM applications have they filed? What percentage were approved but never used (expired)? How many unique roles do they file for? Did the government deny any? Have they filed WARN Act layoff notices while posting jobs? A company that uses most of its approved PERMs scores high. A company with hundreds of expired approvals and active layoff notices scores low.

Final Ghost Score = Text Analysis + Company Track Record

The two scores are blended together. When we can identify the employer in our database, the company's track record carries significant weight - because past behavior is the best predictor of current intent. When no employer is identified, the score relies entirely on text analysis.

What Factors Matter Most

We analyze each posting against PERM filing cross-references, repost frequency decay, WARN Act contradictions, and additional detection signals. Here's how they're grouped by relative importance:

Strongest Impact

A single strong signal here can significantly change the result.

PERM filing cross-reference (same role)

What: The employer filed PERM applications for the same or similar job title.

Why: PERM postings are legally required but not intended to result in outside hires.

Lowers score significantly

WARN Act contradiction

What: The employer filed a federal mass layoff notice within 180 days while posting jobs.

Why: Simultaneously firing and hiring often indicates ghost postings.

Lowers score significantly

Scam indicators

What: Requests upfront payment, unrealistic pay, SSN before hiring, or MLM language.

Why: Hallmarks of employment fraud, not real job postings.

Lowers score to near zero

Strong Impact

Multiple signals in this tier compound.

Company PERM expiry rate

What: Percentage of approved PERMs never used. Why: High expiry = compliance theater.

Higher expiry = lower score

PERM boilerplate language

What: "Foreign equivalent," "suitable combination," mail/fax resume. Why: From PERM regulations, not real recruiting.

Lowers score

Posting age

What: How long the posting has been active. Why: 89% of real positions fill within 30 days.

Older = lower score

Hidden end client (staffing)

What: "Fortune 500 client" without naming the employer. Why: The posting may be speculative.

Lowers score

Salary vs prevailing wage

What: How salary compares to DOL minimum. Why: Pinned at the floor = calculated compliance.

At floor = lowers. Above market = raises

Moderate Impact

Individually small, they compound when multiple appear together.

Description vagueness

Heavy buzzwords ("dynamic," "self-starter," "ninja") without specific duties. Real roles describe concrete work.

Technology stack coherence

15+ unrelated technologies listed is not a real role - no human has this stack.

Contract/staffing language

C2C, W2 contract, 1099, rate DOE. Staffing agency markers - the posting may be speculative.

Sponsorship availability

Informational for visa holders. Does not affect score.

Positive Signals (Raise Score)

These indicators suggest a genuine job opening.

Specific duties and projects

Concrete work, named projects, specific products.

Salary listed and above market

Transparent pay exceeding DOL prevailing wage.

Comprehensive benefits

Health, dental, 401k, equity, PTO listed in detail.

Named team or manager

Real people attached to real openings.

Fresh posting (under 14 days)

Strongly correlates with active, funded hiring.

Interview process described

Active pipeline with defined stages.

What the Numbers Mean

80 - 100

Likely Real

Strong indicators of a genuine opening. Multiple positive signals. Apply with confidence.

60 - 79

Some Concerns

Some yellow flags. May still be real. Research further before investing significant time.

30 - 59

Suspicious

Multiple risk signals. Characteristics of ghost jobs or PERM compliance. Verify directly with the employer.

0 - 29

Probable Ghost / PERM

Strong indicators of a non-genuine posting. Likely does not exist as a real opening.

What Ghost Scores Are NOT

  • Not a statement of fact. Ghost Scores are editorial opinions generated by algorithmic analysis. We do not claim any employer is acting illegally or intentionally deceiving applicants.
  • Not a guarantee. A high score does not guarantee the job is real. A low score does not guarantee it's fake. Scores indicate probability based on available data.
  • Not legal advice. Ghost Scores should not be used as the sole basis for employment decisions, legal claims, or regulatory complaints.
  • Not a judgment of workers. PERM detection identifies employer filing patterns, not the legitimacy of sponsored workers. The workers are not at fault.

Accuracy and Limitations

Ghost Scores are most accurate when the employer is identified and exists in our PERM database. Company track record data provides ground-truth context that text analysis alone cannot.

Text-only scores (when no employer is detected) are less reliable. A well-written ghost posting will score higher than it deserves. A poorly-written real posting will score lower.

Our PERM database updates quarterly when the DOL publishes new disclosure data. Between updates, recent filings may not be reflected in scores.

We continuously improve detection through user feedback, additional data sources, and signal refinement. Scores may change over time as our methodology evolves.

For Employers

If you believe your company's Ghost Factory Score is inaccurate, we welcome disputes. Our scores are based on publicly available DOL data - if the underlying data is incorrect or outdated, we want to know.

Submit a Score Dispute →

© 2026 Tech Cold Brew, LLC. Ghost Scores are editorial opinions based on publicly available data. Not legal or career advice.