Author –Anchal Singh

Insights from SustainoMetric’s Assessment Across 50 Companies

Over the past decade, we have seen ESG and sustainability analysis transform and evolve from a largely manual, judgment-driven discipline into a data-intensive, technology-supported function. What once required analysts to sift through hundreds of pages of sustainability reports, annual filings, and policy documents is now increasingly augmented and, in some cases, accelerated by artificial intelligence.

This shift isn’t just a matter of efficiency; it’s a structural necessity.

The scale of ESG reporting today is unprecedented. A single company can publish disclosures across multiple formats, such as annual reports, sustainability reports, climate disclosures, and governance statements, often running into hundreds of pages. Multiply that across portfolios, and the analytical burden becomes unsustainable without automation.

AI is stepping in to bridge that gap.

From ESG rating agencies to institutional investors, AI is now widely deployed to extract, classify, and structure sustainability data. Tasks that once took days can now be executed in minutes. After doing a small experiment, we realised that AI has made an hour-long task easy, but we cannot also deny the reality that AI is not universally effective; it is highly context dependent.

Which leads to the more important question:
Where does AI perform well in ESG analysis—and why?

Our Approach: Moving Beyond “Does AI Work?”

To solve this question, we at SustainoMetric conducted an internal assessment across 50 publicly listed companies, evaluating AI-assisted ESG analysis against 30 indicators spanning:

• Emissions reporting
• Governance structures
• Policy disclosures
• Workforce metrics
• Supply chain practices
• Sustainability commitments

The goal was not to validate whether AI works – it does.
The objective was to understand where it works reliably, where it adds decision value, and what drives that performance.

The conclusion was both clear and consistent: AI performs best when ESG disclosures are structured, explicit, and measurable.

Where AI Already Delivers Strong Results

In practice, AI’s biggest strength lies in processing and standardizing large volumes of disclosure data quickly and consistently.

Across our assessment, AI showed high reliability in identifying:

• Governance structures and board-level oversight
• ESG-related policies and frameworks
• Climate commitments and sustainability targets
• Quantitative emissions disclosures

These are not coincidental strengths. These disclosures tend to follow repeatable patterns and semi-standardized formats, making them well-suited for machine extraction.

From an operational standpoint, this creates two immediate advantages:

• Scale and Efficiency
  AI enables rapid screening across large datasets, significantly reducing manual workload.
• Consistency
  Unlike human analysts, AI applies the same logic across all companies, minimizing variability in initial assessments.

At its core, AI excels at answering a foundational ESG question:

“Is the disclosure present and can it be extracted reliably?”

The Quantitative Advantage: Why Numbers Win

One of the strongest signals from our analysis was the performance gap between quantitative and narrative disclosures. AI performs significantly better when dealing with numerical, well-defined data. Take emissions reporting as an example. Many companies disclose:

• Scope 1 and Scope 2 emissions
• Reduction targets with baseline years
• Net-zero commitments with defined timelines

These data points are:

• Structured
• Comparable
• Terminologically consistent

As a result, AI systems can extract and interpret them with high accuracy—particularly when reporting boundaries and methodologies are clearly defined. This makes AI especially powerful in climate analytics at scale, where thousands of companies disclose emissions data annually.

Policy and Governance: Structure Enables Detection

AI also demonstrates strong performance in identifying policy and governance disclosures, particularly when they are clearly segmented. Documents such as: Environmental and social policies, Human rights frameworks, and Codes of conduct, etc., are often presented in distinct sections or standalone documents, making detection relatively straightforward.

Similarly, governance elements like Board oversight of sustainability, ESG committees, and Executive accountability tend to follow recognizable disclosure patterns across companies.

This allows AI to systematically identify governance structures, enabling faster benchmarking and cross-company comparisons.

Commitments and Targets: Clarity Drives Accuracy

Another area where AI performs reliably is in capturing well-defined sustainability commitments.

Targets that include:

• Specific percentage reductions
• Clear timelines
• Defined baselines

are consistently interpreted with higher accuracy.

For example:
A commitment to reduce Scope 1 and 2 emissions by 30% by 2030 from a 2020 baseline provides the exact structure AI systems are designed to process.

The broader takeaway is critical: The more precise and measurable the disclosure, the more dependable the AI output.

The Core Insight: Structure Is the Enabler

Across all ESG themes, one pattern holds:

AI performance is directly proportional to the level of structure in the disclosure.

AI delivers the strongest results in areas such as:

• Emissions data
• Quantified targets
• Formalized policies
• Governance frameworks

In these domains, it materially improves:

• Speed of analysis
• Consistency across datasets
• Scalability of ESG evaluations

For investors, analysts, and sustainability teams, this is not just an efficiency gain; it’s a capability upgrade.

But ESG Is Not Always Structured

However, after years of working with ESG data, it’s equally clear that the most critical insights are often the least structured.

Key risk areas such as:

• Supply chain practices
• Workforce conditions
• Implementation of policies
• Quality of governance

are frequently embedded in narrative, context-heavy disclosures.

These require:

• Interpretation
• Contextual judgment
• Cross-referencing

Capabilities where AI still has meaningful limitations.

And this is precisely where over-reliance on automation can introduce risk.

What Comes Next

This analysis highlights where AI is already delivering tangible value, a strength, but it also sets up the more important discussion.

In the next article, we will examine:

👉 Where AI struggles in ESG analysis and why those gaps matter

From supply chain transparency to workforce disclosures and governance interpretation, these areas expose the limits of automation and the continued importance of human expertise.

Because the future of ESG analysis is not AI versus humans.

It is AI working alongside human judgment, each applied where it performs best.