An ESG score is a number that summarises an organisation’s performance on a selected set of environmental, social and governance (ESG) indicators, calculated under a stated methodology. A score is only as meaningful as that methodology: which indicators it includes, how they are weighted, how missing data is treated and which organisations it is compared with.
This guide answers the questions sustainability, finance and procurement teams most often ask about ESG scores, and sets out the governance that makes a score trustworthy.
What is an ESG score?
An ESG score condenses many sustainability data points (for example emissions intensity, water use, safety performance, workforce diversity or board independence) into one figure, often with separate sub-scores for the environmental, social and governance pillars.
Scores are used in three common ways: inside a company, to track sustainability performance over time; by investors and rating providers, to compare companies; and by buyers, to assess suppliers. The same word can therefore describe very different calculations, which is why the methodology behind a score matters more than the number itself.
How is an ESG score calculated?
The details differ between methodologies, but most ESG scores follow the same five steps:
- Select indicators. Choose the metrics that are material for the organisation or its sector, such as Scope 1 and 2 emissions intensity, water withdrawal, lost-time injury frequency rate (LTIFR) or the share of independent directors.
- Normalise. Put indicators on a common scale, for example by comparing each value with a peer group, a target or a threshold, so that tonnes of CO₂e and percentages can be combined.
- Weight. Decide how much each indicator, and each E, S and G pillar, contributes. Weights often reflect materiality, so they can differ by sector.
- Aggregate. Combine the weighted indicator results into pillar scores and an overall score.
- Treat missing data. State how gaps are handled (excluded, estimated or penalised), because this choice alone can move a score significantly.
A transparent methodology documents each of these steps, together with the scoring scale, the reporting period and a version history, so that a change in method is never mistaken for a change in performance.
What factors influence a company’s ESG score?
The indicators that are material to the business carry the most weight. Typical examples include:
- Environmental: Scope 1, 2 and 3 greenhouse gas emissions and their intensity, energy use and renewable share, water withdrawal and discharge, waste generated and diverted from landfill.
- Social: health and safety, employee wellbeing and development, diversity and inclusion, human rights in the value chain and community impact.
- Governance: board composition and independence, ethics and anti-corruption, risk management and the quality of sustainability disclosures.
Data coverage also matters: an organisation that cannot evidence a metric may score lower than one that can, even when actual performance is similar. A double materiality assessment is a structured way to decide which topics should carry the most weight.
What is the difference between an ESG score and an ESG rating?
The terms are often used interchangeably, and their meaning varies by provider. A useful distinction is that an ESG score is the numeric result of a defined calculation, while an ESG rating is usually an assessment published by a third-party provider on its own scale (letters, numbers or risk categories), based on its own research, data sources and methodology.
Because providers choose different indicators, weights and data, the same company can receive noticeably different ratings. Academic research has documented this divergence; see Berg, Kölbel and Rigobon, “Aggregate Confusion: The Divergence of ESG Ratings”, Review of Finance (2022). An internal performance score calculated from a company’s own data is not an independent third-party rating and should not be presented as one.
How can a company improve its ESG score?
By improving the underlying performance and the quality of the evidence behind it, not the score itself. In practice:
- Find the gaps that matter. Compare each indicator with targets or peers and rank the gaps by their effect on the score.
- Assign owners and targets. Give every gap an accountable owner, a target value and a deadline.
- Act and track. Turn gaps into corrective actions with milestones, and review progress every reporting period.
- Close data gaps. Replace estimates with measured, evidenced values where you can.
In SustainStep, Benchmarking compares your sustainability metrics with peers in your industry segment, and its gap analysis ranks the metrics that would move your sustainability rating the most. GHG Accounting models decarbonisation levers and tracks progress against targets.
Can AI help calculate and analyze ESG scores?
Yes, within limits. AI is useful for the work around a score: reading source documents, catching anomalies and missing values, mapping metrics to frameworks and explaining what drives a result. It should not replace a documented methodology or accountable human review.
The SustainStep AI Engine extracts data from bills, invoices and certificates, flags outliers and missing values as data arrives, and answers plain-language questions with the source records it used; drafted disclosure text is reviewed by people before it is published. Whatever tools you use, keep every AI-assisted figure traceable to its evidence.
How can organizations compare ESG scores across suppliers?
Only when every supplier is assessed on the same criteria, with the same evidence rules and in a comparable context. Use one questionnaire per supplier category, define which evidence is accepted, score responses against stated criteria, and read results in sector context: a logistics provider and a chemicals supplier face different material issues.
SustainStep Value Chain Assessment sends sustainability questionnaires through a supplier portal, scores responses against your criteria and sector benchmarks, and flags high-risk suppliers by category, geography and spend.
How often should ESG performance scores be updated?
As often as the underlying data changes reliably, and at least once per reporting period. Indicators fed by monthly data, such as energy, fuel, water and waste, can be tracked monthly or quarterly; others, such as board composition or annual survey results, change once a year. Every score should state the period it covers, and scores calculated under different methodology versions should not be compared without restating them.
How should missing ESG data be treated?
Transparently. Label every input as verified, estimated, missing or not applicable, and never treat these states as equivalent:
- A not applicable indicator (for example refrigerant leakage at a site with no refrigeration) should not lower a score.
- A missing value should not be silently counted as zero or as average performance.
- An estimate should state its method and be replaced with measured data when it becomes available.
Report data quality separately from performance, so that better evidence is not mistaken for better performance. In SustainStep Data Management, every value is linked to its source document, approver and timestamp.
Can a company’s ESG score be compared across industries?
Only with caution. Material issues differ by sector (emissions intensity matters more for cement than for software, data privacy more for financial services), so scores built on sector-specific indicators and weights are not directly comparable across industries. Meaningful comparison needs a defined peer group (sector, size, geography and period) and a consistent normalisation method. Cross-industry comparisons are best limited to indicators every sector shares, and presented with their limitations.
SustainStep Benchmarking builds peer groups by sector, size and geography so that each comparison is like-for-like.
Checklist: governance for a trustworthy ESG score
- Methodology: document criteria, weights, scale, aggregation and version history.
- Data provenance: keep the source, reporting period, reviewer and evidence link for every scored input.
- Missing data: mark missing, estimated, not applicable and verified inputs explicitly.
- Benchmark comparability: define the peer group, sector, geography, period and normalisation method.
- Score versus rating: do not describe an internal score as an independent third-party rating.
- Claims: avoid unsupported claims of certification, regulatory endorsement or universal comparability.
Related SustainStep capabilities: Value Chain Assessment, Benchmarking and SustainStep AI Engine.