What is anomaly detection in ESG data?

ESG anomaly detection identifies unusual changes or patterns in a company’s ESG scores when compared to its own historical performance and to peer companies within the same industry.

Rather than treating all score movements as equal, anomalies are characterised by:

• Magnitude — how unusual the change is

• Polarity — whether the change reflects a significant improvement or deterioration

This helps users distinguish between normal variation and signals that genuinely require attention — such as potential data errors, structural shifts, or greenwashing risks.

Who is this for?

Designing trust, clarity, and action in complex sustainability data

Regulators

• Assess market readiness for ESG regulations

• Augment investigation and supervision workflows

Corporates

• Improve quality, consistency, and completeness of ESG disclosures

• Understand performance relative to industry peers

• Strengthen internal target-setting and reporting credibility

Asset Managers

• Improve portfolio quality through sharper ESG risk analysis

• Identify material risks and opportunities earlier

• Generate alpha through more informed portfolio adjustments

Assurance

• Identify areas requiring scrutiny

• Focus advisory efforts where they add the most value

How it works

The framework combines statistical analysis, machine learning, and explainability techniques to surface actionable ESG intelligence

At a high level:

• ESG category-level score changes are calculated year-over-year

• Changes are normalised and benchmarked within industry peer groups

• Statistical techniques (e.g. z-scores) identify unusual deviations

• Machine learning models trained on 18+ years of data across ~10,000 companies detect subtle outliers

• Model explainability tools quantify which ESG categories contribute most to each anomaly

• Results are translated into clear magnitude and polarity signals that are easy to interpret

This allows users to move beyond headline ESG scores and understand what changed, how unusual it is, and why.

How it’s applied in practice

• Each dot represents a company

• Position reflects anomaly magnitude and polarity

• Dot size represents portfolio weight

• Colour indicates industry sector

Most holdings cluster near the centre, indicating low anomaly risk.
Outliers — especially those with large magnitude — immediately stand out and prompt review.

From there, users can:

• Drill down into contributing ESG categories

• Compare against peer benchmarks

• Decide whether to investigate, rebalance, or engage

Anomaly insights are surfaced through intuitive visualisations such as portfolio scatter plots:

Role

• Explainability: Designing clear signals that show why something was flagged and how it compares to peers

• Clarity of severity: Visual hierarchy to separate critical anomalies from mild deviations

• Traceability: Drill-down paths linking anomalies back to underlying disclosures and categories

• Comparability: Peer-benchmark views that make deviations instantly understandable

• Workflow integration: Ensuring anomaly review fit naturally into ESG analysis, reporting, and due-diligence flows

Portfolio Use case

" Its Important to see the reason behind the score changes- Including inconsitencies in ESG data"- Portfolio Manager USA

" Competiors are building models to show how risky my portfolio is in terms of Carbon Emissions"

What’s next

The roadmap extends anomaly detection beyond scores alone by:

• Linking anomalies to news, events, and emerging trends

• Deepening explainability at sub-category and metric levels

• Supporting more proactive alerts and scenario analysis

This evolution continues to strengthen confidence in ESG data — helping users not just detect anomalies, but act on them with clarity.