ATS Research Project

A comprehensive research project analyzing Applicant Tracking System (ATS) parsing capabilities and optimization techniques for CV/resume processing.

UPDATE: Suspended as of 22/09/2025 for lack of meaningful results, most of the modern ATS systems use specific filters for 90% of common "CV hacks", but i will look forward to strenghten the bypasses someday.

Overview

This project investigates how different ATS systems parse and extract information from CVs in various formats, with a focus on understanding the effectiveness of different optimization strategies.

Project Structure

ATS-Research/
├── CVs/                    # CV variants (HTML source files - not tracked)
├── PDFs/                   # Generated PDF outputs for testing
├── analysis/               # Data analysis and research notebooks
├── convert-to-pdf.bat      # Windows conversion script
└── convert-to-pdf.sh       # Unix/Linux conversion script

CV Variants

The research includes four distinct CV versions designed to test different scenarios:

• Control: Baseline CV with standard formatting
• Minimal Control: Simplified version with minimal styling
• Experimental: Enhanced with ATS optimization techniques
• Experimental Stealth: Advanced version with aggressive ATS hacks

Data Collection Methodology

Research Design

This study employs a controlled experiment comparing four CV variants across multiple job applications to measure ATS parsing effectiveness and response rates.

CV Variants & Testing Strategy

1. Control: Standard professional CV with conventional formatting
2. Minimal Control: Simplified version to test baseline readability
3. Experimental (Hack): Enhanced with standard ATS optimization techniques
4. Experimental Stealth: Advanced version with aggressive ATS manipulation

Data Collection Process

1. Job Application Workflow

Job Posting → CV Selection → Application Submission → Response Tracking → Data Logging

2. Data Points Collected

For each job application, the following metrics are recorded:

Application Details:

• Company name and role
• Platform used (LinkedIn, Indeed, company website, etc.)
• Date of submission
• CV variant used (Control/Minimal/Hack/Hack Stealth)

ATS Analysis:

• ATS system detected (Workday, Greenhouse, BambooHR, etc.)
• Platform-specific indicators
• Parsing behavior observations

Response Tracking:

• Response received (Yes/No)
• Response time (days)
• Response type (automated rejection, human contact, interview invite)
• CV mentioned in response
• Specific feedback on CV format
• Red flags detected in communication

3. Automated Data Collection

Google Forms Integration:

• Standardized data entry form for consistent logging
• Automatic timestamping
• Real-time alerts for positive responses to optimized CVs

Response Rate Monitoring:

• Weekly automated reports comparing variant performance
• Platform effectiveness analysis
• ATS system compatibility scoring

ATS Optimization Techniques Tested

Standard Techniques (Hack Version):

1. Hidden Keywords: Strategically placed invisible text using CSS positioning
2. Semantic HTML: Structured markup for better parsing
3. Microtext Elements: Ultra-small font content with relevant keywords
4. Metadata Embedding: Custom data attributes and meta tags

Advanced Techniques (Hack Stealth Version):

5. Unicode Manipulation: Zero-width characters and special encoding
6. SVG Hidden Text: Vector-based invisible content
7. Canvas Rendering: Programmatic text insertion
8. CSS Pseudo-elements: Generated content techniques
9. PDF-Specific Layers: Techniques that survive HTML-to-PDF conversion

PDF Generation & Preservation

Conversion Process: CVs are converted from HTML to PDF using wkhtmltopdf with optimized parameters specifically designed to preserve ATS-friendly elements.

Critical Parameters:

--page-size A4
--margin-top 0.5in
--dpi 300
--enable-javascript
--background
--no-pdf-compression

ATS Hack Preservation: Special CSS rules ensure hidden elements remain parseable in PDF:

@media print {
  .ats-hidden {
    position: absolute !important;
    left: -9999px !important;
    opacity: 0.001 !important;
    font-size: 1pt !important;
  }
}

Statistical Analysis Plan

Primary Metrics:

• Response Rate by CV Type: Percentage of applications receiving responses
• Response Quality: Type and speed of responses
• Platform Performance: Effectiveness across different job boards

Secondary Metrics:

• ATS Detection Rate: How often systems identify optimized elements
• Parsing Accuracy: Completeness of information extraction
• Human Readability Impact: Correlation between optimization and presentation

Sample Size Considerations:

• Minimum 20 applications per CV variant for statistical significance
• Stratified sampling across industries and company sizes
• Control for role similarity and application timing

Research Objectives

Primary Hypotheses:

1. ATS-optimized CVs will show higher response rates compared to standard formats
2. Advanced stealth techniques will outperform standard optimization while maintaining visual quality
3. Different ATS systems will show varying sensitivity to optimization techniques

Measurement Goals:

• Quantify ATS optimization effectiveness across real-world applications
• Identify optimal techniques for different platforms and ATS systems
• Balance optimization impact with human readability requirements
• Document ATS parsing behaviors and system-specific preferences

Usage

Generate PDFs

Windows:

cd CVs
.\convert-to-pdf.bat

Unix/Linux:

cd CVs
chmod +x convert-to-pdf.sh
./convert-to-pdf.sh

Analysis

Research data and analysis can be found in the analysis/ directory using Jupyter notebooks.

Requirements

• wkhtmltopdf (for PDF generation)
• Python 3.x (for analysis)
• Jupyter Notebook (for research documentation)

Expected Results & Analysis

Data Visualization

Research data and analysis can be found in the analysis/ directory using Jupyter notebooks, featuring:

• Response rate comparisons across CV variants
• Platform effectiveness heatmaps
• ATS system compatibility matrices
• Time-series analysis of application success

Key Research Questions

1. Do ATS optimization techniques significantly improve response rates?
2. Which specific techniques are most effective across different ATS systems?
3. How do different job platforms respond to optimized CVs?
4. What is the trade-off between ATS optimization and human readability?

Statistical Significance

• Minimum sample size: 80 applications (20 per variant)
• Confidence level: 95%
• Expected effect size: 15-25% improvement in response rates
• Control variables: Role similarity, industry, company size, application timing

Data Collection Status

PDF outputs are automatically saved to the PDFs/ directory, organized for systematic ATS testing across different platforms and systems.

Current Status: 🟡 Data Collection Phase

• ✅ CV variants finalized and optimized
• ✅ Automated tracking system configured
• ✅ Google Forms integration active
• 🔄 Applications in progress
• ⏳ Statistical analysis pending sufficient sample size

Ethical Considerations

Research Ethics

• Transparency: This is legitimate A/B testing of CV formats for research purposes
• No Deception: All CV content is truthful and accurate
• Professional Standards: Optimization techniques enhance rather than falsify qualifications
• Educational Value: Results will contribute to understanding of ATS systems

Responsible Disclosure

• Research findings will be shared to benefit the broader job-seeking community
• No proprietary ATS vulnerabilities will be exploited
• Focus on improving accessibility and fairness in hiring processes

---

This is a research project for educational and professional development purposes. All applications contain accurate information and are submitted in good faith.