Business Situation
In a rapidly evolving ESG landscape, the client a global sustainability ratings provider needed to perform deep, scalable environmental, social, and governance (ESG) risk assessment across its coverage universe of over Top 600 stocks from Index worldwide.
Their goals were threefold:
1. Detect company involvement in controversial sectors or behaviors (e.g., arms, tobacco, pollution, governance lapses) through rigorous product screening and controversy analysis
2. Evaluate alignment with the 17 UN Sustainable Development Goals (SDGs), both directly and indirectly
3. Ensure data quality, transparency, and consistency to empower investor decisions and maintain credibility as a rating agency
Given the complexity and volume of data, they sought a trusted partner with a dual advantage: technology-powered workflows and domain-specific ESG expertise.
SGA Approach
SG Analytics designed a hybrid engagement that seamlessly integrated artificial intelligence (AI)-driven data validation, natural language processing (NLP) pipelines, and a deep pool of ESG domain experts across 16 sectors and 40 sub sectors.
This fusion of technology and human intelligence ensured scalability, granularity, and defensibility of ESG and SDG-related insights.
Our Tech-Driven ESG Research Framework
1. Custom ESG Criteria Definition & Screening Logic
- Partnered with the client to define ESG negative screening thresholds across sectors
- Mapped revenue thresholds and qualitative risk markers across activities including tobacco, weapons, fossil fuels, gambling, and more
- Designed a rule-based AI engine to automatically flag disclosures indicating non-compliance with these thresholds
2. Multi-Layered Controversy Detection Engine
- Built a hybrid controversy analysis framework combining:
– AI-powered web scrapers and NLP classifiers to scan thousands of news reports, regulatory filings, and media updates
– SGA’s internal Controversy Knowledge Graph, mapping over 45 controversy indicators to categories such as pollution, anti-competitive behavior, human rights violations, and corporate governance lapses - Our specialized team of analysts validated, contextualized, and scored each flagged controversy based on credibility, severity, and recurrence
3. Proprietary AI Validation Modules (IDEAT-QA & Sentiment Verifier)
- Integrated IDEAT-QA for automated detection of inconsistencies across disclosures, data duplication, and source credibility scoring
- Deployed a Sentiment Verifier model to assess public and stakeholder sentiment around controversies for reputational risk scoring
4. SDG Mapping and Contribution Analysis
- Developed a decision-tree framework to sector-based company activities with SDGs, using industry-specific context and sustainability taxonomies
- Classified contributions as Direct, Indirect, or Neutral across each SDG
- Created sector-weighted impact profiles to benchmark companies’ alignment within peer groups
5. Domain-Expert Analyst Oversight
- A team of 50 ESG specialists with sectoral specialization conducted:
– Source triangulation for flagged issues
– Disclosure evaluation for context and credibility
– Quality control checks aligned with client methodology - Analysts regularly updated findings through an ongoing monitoring model, ensuring ESG positioning remained current and relevant
Key Takeaways
ESG Expertise Meets AI Automation
The engagement’s strength came from integrating AI models for high-speed screening and NLP analysis with subject-matter ESG judgment for credibility, nuance, and context achieving 98% accuracy
Structured & Defensible Methodology
SOPs and pre-defined frameworks ensured transparency, replicability, and alignment with leading ESG standards (e.g., UN SDGs, SASB, and GRI)
Future-Proofed Research Workflows
The deployment of internal AI validation modules (IDEAT-QA, Sentiment Verifier) reduced manual rework by 35%, increased consistency, and enabled seamless scaling
Credibility Anchored in Domain Expertise
ESG analysts brought essential contextual intelligence, interpreting nuanced disclosures, validating 600 companies, 750 impact outcomes aligned to UN SDGs
