When our founder and CEO, Charles Radclyffe, wrote this piece for the Financial Times during 2020 – he saw an opportunity to better assess companies’ ESG alignment and maturity using AI. In fact, when he was later interviewed by a journalist at ESG Investor about what he saw as the risks of using AI in the context of ESG – he said “companies using AI to write ESG reports is a particularly scary prospect – given the risk of gaming the system”! Now years later, he is running EA, a software company with an AI platform designed to help companies automate diligence processes, such as those instigated by ESG ratings agencies like EcoVadis, MSCI, and others – how do we feel about AI and ESG?
AI to analyse ESG performance
First, let’s explore the use of AI to analyse ESG performance:
There is nothing new about the use of AI to analyse companies’ results. For as long as advanced data analytics techniques have been available, ratings agencies, proxy advisors, and asset managers have been building and buying tools to help them analyse sustainability reports, disclosures and other credentials.
EcoRatings claimed to have launched the first AI-powered ESG platform at COP28 in 2023; although the world’s largest sustainability-focussed advisory firm, ERM, made a similar claim as early as February 2022. Although it’s hard to say who was the first to market, the title for first AI driven ESG assessment solution is likely to be held by Clarity AI, who launched their first product publicly in 2019.
So, how is AI used in such a context? Really this boils down to two methods – the use of statistical analysis to evaluate companies’ various reported metrics (such as gender pay gap, or CO2 per $m revenue) and benchmark these against competitors; and the use of natural language processing systems to ‘read-between-the-lines’ either in companies’ corporate communications.
Beyond this, there is also the challenge of analysing any correlations between ESG datasets such as that provided by EcoVadis, CDP, Sustainalytics or other scoring methodologies and data made available in the public domain.
The purpose? An ESG score is by itself a useless measure, unless translated into a $ value. For an investor, it’s essentially a question of how likely particular risks might surface, and what the cost of dealing with such risks might be (both operational expense and capital value loss). In supply-chain analysis, it’s a similar question – and usually driven by a risk the customer is trying to mitigate (such as the risk of a ransomware attack that might cause key IT infrastructure to go offline).
AI to streamline ESG disclosure
At EA, we’re solving a very different problem to the ESG ratings agencies. While they are trying to provide better insight to investors or those trying to understand their supply-chains, we are in the business of helping those on the receiving end of these requests deal with them more efficiently.
Firstly, let’s deal with the elephant in the room – should AI be used at all in this context? The answer to this is an emphatic ‘yes’!
To explain why, let’s evaluate the problems with handling ESG disclosure in ‘the old-fashioned-way’.
Manual ESG disclosure is time-consuming
Anyone who has been on the receiving end of an ESG disclosure request will know how time-consuming the activity is if completed manually.
First a spreadsheet needs to be created which contains the request in full; a row for each question, and a column for each data point required (answer, evidence, verbatim text, page number, etc).
Next, this spreadsheet needs to be filled-out. Each cell in the spreadsheet will contain some text once completed, each typed out or copy-and-pasted in. Some answers can be done from wrote memory. Others will need scouring through documents to find the relevant answer. And still more will require constant emails, DMs, and telephone calls with colleagues to chase down the latest, up-to-date information.
Manual ESG disclosure is error-prone
All this manual work inevitably leads to mistakes being made. This is the single biggest reason why EcoVadis reporting prepared by AI can score more highly than that written by humans – simply because it contains no errors.
One of our long-standing customers astonished us recently when they admitted that even at the end of the approval process which included risk, legal, corporate comms, and compliance colleagues – the Chief Sustainability Officer himself sat down before a CDP submission was made and manually checked each entry line by line against the master spreadsheet in case an error was made in the final copy-and-paste.
AI solutions like EA bring an end to this madness, and give confidence to all involved that every answer is accurate, up-to-date, and consistent.
Manual ESG disclosure is dull work
If you’re reading this and recognise the task we’re describing, you’ll know first hand how boring it is. This is not people’s finest work. It’s not what we were put on this planet to deliver. And crucially, this is not the work that people drawn to sustainability roles were inspired to deliver.
A Chief Sustainability Officer we recently spoke with confessed that she turned down an ESG ratings agency recently in the week before a family holiday. On the first day of her vacation, she got a call from the CEO pleading with her to complete it, as a key customer depended on the metric being positive, and it could influence a sales deal. There she was, with the family by the pool, holed up in her hotel room filling in a form. Everyone knows it was not her best work, and yet – she diligently got it completed on time.
That same afternoon, as she was sitting by the pool she did a Google search for “AI to automate ESG ratings” and found EA. We were one of her first calls when she returned from vacation – having vowed never again to interrupt a family holiday for ESG reporting!
Manual ESG disclosure is distracting to driving impact
Another Head of ESG Reporting had his two senior analysts quit after the 2024 CDP submission because the workload had been unbearable. When top talent who are motivated to make their companies better organisations quit because of form-filling – the situation is clearly unsustainable. This is exactly why we built EA to solve!
Conclusion
With platforms like EA’s Easy Autofill solution for ESG reporting, ESG disclosure requests can be responded to in a matter of a few clicks. EcoVadis reporting, as an example can be reduced to under 90 minutes end-to-end; and that includes the production of the first draft as well as the entire review, edit, and approval process. Not only does this save time for all involved, freeing them from dull work, reducing the risk of key staff attrition; but also this leads to fewer errors, better communication of data points which need to be highlighted, and ultimately – allows human time to be repurposed on devising and execution of strategy – leading to positive impact.