The Emerging Role of Natural Language Processing in ESG Analytics

esg analytics

Artificial intelligence, and in particular natural language processing (NLP) could enable investors to efficiently access and analyse data and “dramatically increase” the scope of analysis to assess environmental, social, and governance (ESG). This is according to a joint paper by the International Finance Corporation and asset management company, Amundi, earlier this year. 

Why is ESG Analysis Important?

In light of recent and ever-more-frequently occurring world events, environmental, social, and governance, or ESG, is taking on an ever-greater significance. There is a global trend for responsible investing as consumers, particularly those from younger generations, are creating a demand for responsible investments.

This trend was prominently illustrated earlier this year when Cardano (touted as one of the most environmentally-friendly cryptocurrencies) defied a major downturn in the industry after Elon Musk tweeted concerns about the environmental impact of the crypto mining process.

As well as futureproofing your investment, analysing ESG is also the socially responsible path. Fraud, corruption and climate change all pose a systemic risk for society.

The Challenges of ESG Analytics

Investors are frustrated by the lack of standardised ESG data. The growth of the market has led to an increase in ESG data and rating providers, but that has only added to the confusion in the market, experts said.

The global ESG fund has tripled to over one trillion dollars since 2015. This growth has led to the fast emergence of ESG rating providers, each with different metrics and parameters. Issuers are bombarded with forms leading to what the IFC/Amundi report dubs “ESG reporting fatigue”.

The rapid growth and the resultant increase in awareness of the importance of ESG data has almost entirely been within the developed world. This makes it highly challenging for investors to assess ESG and invest in issuers in emerging markets.

Another challenge is the complexity of such a multidimensional field. How do you compare a company’s anti-corruption policies with its commitment to sustainability? How do you make a like for like assessment of fraud prevention strategies from one region to another? Structured data alone, which is subject to discrepancies, won’t give you a three-dimensional picture in such a nuanced field.

Quality and traceability are also challenges for ESG data. Analysts need access to raw data to get a clear and honest understanding of a target company’s performance and commitment to best practices and addressing key issues. 

In summary, lack of regulation along with the complexity of ESG makes the comparison between regions near impossible. But thankfully, emerging capabilities in AI and NLP technology have the potential to provide new data points and new perspectives to existing data to help bridge the ESG disclosure gap.

Use of Unstructured Data and NLP

Investors are using AI technologies to collect and analyse unstructured data to add breadth and depth to ESG assessment.

News articles, project disclosures to multilateral development banks (MDBs), sustainability reports and bond prospectuses are all listed as underused sources of unstructured data. Social media and review data can also support metrics such as employee satisfaction and worker rights.

Historically assessing long-form text has been a time and resource-intensive process. However, innovations in NLP have revolutionised  how unstructured text can be collected, analysed and interpreted.

NLP tools, such as those created by Symanto, can analyse unstructured data rapidly and on a massive scale. Unstructured text can be taken from public sources, eliminating ESG reporting as an absolute requirement, and enabling a more thorough analysis of ESG performance within emerging markets.


By analysing text from media and other document types not specifically created for ESG reporting, there’s a good chance that there will be fewer discrepancies and less likelihood that the data has been manipulated or skewed to give a favourable view of the issuer.


Applications of NLP on ESG Reporting


Expedite manual analysis

NLP can reduce the time it takes to find relevant ESG information within reports and other internal and external documents. For example, UK high street food and clothing retailer Marks and Spencer published this in their sustainability plan:

“Over the past twelve months, we’ve improved our approach to human rights significantly, culminating in the recent Corporate Human Rights Benchmark scoring us as the best apparel and food business.”

Excerpt from Marks and Spencer Plan A 2025 commitments


Instead of manually reading and analysing the report, the Symanto NLP model would perform text classification and sentiment analysis on the report to extract only the most relevant excerpts. In this instance, the Symanto model would classify the passage above as relating to “Human Rights” with a “positive” sentiment value.

NLP can then be used on other sources such as news reports, social media and blogs to either corroborate or disprove promises made in corporate reports.

Measure workplace sentiment/employee satisfaction

Employee satisfaction is a key metric for ESG analysis. How employers treat their workforce isn’t only a measure of their commitment to ethical practices, there is also evidence suggesting that companies with more contented employees also perform better on the stock market.

Get Started with Insights from Symanto

We create solutions to match your exact needs. To find out more about Symanto’s NLP tools, get in touch or book your free personalised demonstration today.