ESG Data Firm Util Leverages NLP for Fresh Approach

By Sanford Bragg February 3, 2021

Artificial intelligence is transforming ESG research, and London-based Util represents one example of the innovative ways AI is being applied.  Using natural language processing, Util evaluates companies as an aggregation of their product revenues and, rather than relying on analyst opinions, the firm draws on peer-reviewed academic journals.

Util seeks to measure the positive and negative ESG impacts of a company’s products and services.  Its methodology quantifies the degree that a company’s revenues positively or negatively impact the 17 United Nations Sustainable Development Goals. The firm covers 50,000 publicly traded companies, up from 1,800 two years ago.   

Util’s analytic process starts with a detailed breakdown of company revenues.  Its technology “reads” over 120 million peer-reviewed academic sources to find connections between a given company’s products and sustainability concepts, then aggregates the positive and negative relationships it finds.  Company revenues are evaluated relative to 2,000 sustainability concepts which are then mapped the 17 UN SDGs and their 169 sub-targets.

Util bases its criteria on academic literature rather than company disclosures or social media sentiment.  “Academic literature is a vast, credible and regularly updated dataset,” said CEO Patrick Wood Uribe in an interview. “It allows us to quantify the impact of thousands of companies without relying on biased sources such as company disclosures or noisy sentiment-driven datasets.”

Founded in 2017 as Effective Investing, the company rebranded and recruited Wood Ulribe as CEO in August 2020.  The new CEO was previously the sales head for Kensho Technologies, which was acquired by S&P Global in 2018. The company currently has 8 employees and is recruiting for data engineer whose role will be to help ingest new data sources according to the job description.  Util is privately funded, having raised $1.4 million in seed capital according to Crunchbase.  

Our Take

Perhaps the most profound change to the ESG data landscape is the growing use of NLP to improve the collection of traditional ESG factors but also to create new categories of data.  The poster child for AI-driven ESG analysis is Truvalue Labs, recently acquired by FactSet, which uses natural language processing to derive ESG sentiment scores based on over 100,000 sources including global, national and local media, NGOs, government agencies, as well as company disclosures. RepRisk, a venerable twenty-two year old firm, revamped its platform last year, allowing it to process over 500,000 risk incidents on a daily basis from 90,000+ news and social media sources using NLP in combination with human curation.  

As a more recent startup, Util’s effort is currently more modest but it benefits from a differentiated approach.  Aligning its metrics with company revenues has intuitive appeal for capital markets obsessing over the latest earnings release.  Its current reliance on academic literature provides credibility, but we would not be surprised if the firm ultimately expands its inputs include company disclosures and news media, following other firms in the AI-driven ESG space.

Related Articles

  1. Robust Growth of MSCI ESG Research Validates Ongoing Investment (12)
  2. ESG Data is Inconsistent, Confusing, Opaque Yet Increasingly Vital (12)
  3. Mergers and AI are Curing ESG Data’s Ills (12)
  4. AI-Driven ESG Provider Clarity AI Expands US Footprint (12)
  5. Has Alternative Data Become More Affordable? (9)
  6. Euromoney Seeks to Stem Declines at BCA and NDR (9)
  7. Is COVID-19 Spurring Alternative Data Growth? (9)