Visualizing the Effects of ESG Management: An Attempt to Unravel Complex Issues with AI (Part 2)

As the demand for ESG (Environment, Social, Governance) management grows, one challenge companies face when considering its adoption is the difficulty in seeing its effects. To address this, Dentsu Inc. International Information Services Co., Ltd. (ISID) Open Innovation Lab (InnoLab), which develops corporate systems, and ITID Co., Ltd., which provides corporate consulting services, are conducting joint research to analyze and visualize the effects of ESG management using AI. In the second part of our interview with Kuniharu Sakai, Director of InnoLab; Fumiichi Matsuyama, Senior Consultant; and Jun Kanie, who was Unit Director of ITID's R&CD Unit at the time of the interview, we discussed the current research findings and future prospects.

Data analysis enables ESG management strategies tailored to specific target audiences

Q. I understand you presented the results of this joint research at an event held in 2022. What were the key findings?

Matsuyama: Under the theme "ESG Leading to Tangible Results: Unraveled Through Data Analysis – For Capital Markets / Goods Markets / Labor Markets," we analyzed ESG initiatives by dividing them into three target groups: "Capital Markets = Impact on Investors," "Goods Markets = Impact on Consumers," and "Labor Markets = Impact on Job Seekers." In other words, we clearly analyzed and demonstrated "what kind of messaging is effective for whom."

Matsuyama: The analysis involved three steps. First, we used ISID's "Causal Analysis via CALC(※)" to extract ESG themes with clear causal relationships. Next, we employed ITID's "Predictive Analysis via Machine Learning" to identify ESG themes where impact prediction is feasible. Finally, we combined these two results to extract themes with strong causal relationships and high predictive validity.

Matsuyama: The analysis utilized publicly available corporate ESG data (approximately 400 items). For example, E (Environment) included greenhouse gases (CO2) and biodiversity; S (Social) covered diversity and respect for human rights; G (Governance) encompassed compliance and governance. We also used corporate financial data and image data.

The analysis results indicate that themes effectively influencing investment (PBR) include "Disclosure of Director Information" under G (Governance), "Reduction of CO2 Emissions" under E (Environment), and "Establishment of Quality Management Systems" under S (Social).

Themes effectively influencing employment intent included "reducing CO2 emissions," "reducing toxic chemicals," and "efficient energy use" under E (Environment), and "contributing to customer health" under S (Social).

Furthermore, themes found to effectively influence purchasing intent include "Energy Efficiency," "CO2 Emission Reduction," and "Effective Use of Water Resources" under E (Environment).

Summarizing these results into the four themes with the highest priority for initiatives by target group yields the table below.

Matsuyama: In short, it suggests that for investors, ESG should be communicated in a balanced manner, while for job seekers and consumers, the E (environmental) aspect should be actively emphasized. Regarding the overall prevalence of E items and the relative scarcity of S (social) and G (governance) items, one contributing factor may be that, compared to overseas companies, Japanese companies lack a sufficiently developed culture of disclosing information in these areas.

Kanie: This initiative presents general principles applicable across all industries. We aim to advance more specialized analysis tailored to specific sectors and individual companies to enhance the utility of these findings. Furthermore, as ESG trends evolve, we plan to update our analysis annually to regularly assess which factors contribute to enhancing corporate value.

Visualizing effects through data analysis drives corporate ESG adoption

Q. It was surprising to learn that effective ESG elements differ by target audience. For example, knowing that focusing on a specific item could help attract more job applicants would be highly useful for corporate strategy.

Kanie: That's right. While "ESG management" might sound like it's only about headquarters departments, it's fundamentally an activity that should involve the entire company, including frontline operations. This survey likely helped identify recommended activity themes for each department when they were struggling to select themes.

Based on these findings, departments like finance or corporate planning that interact with investors could see benefits by broadly understanding and communicating the company's overall ESG themes. Furthermore, HR departments and internship-related units, which engage with job applicants, could enhance applicant interest by communicating not only HR-related themes like well-being and engagement, but also information on "CO2 emissions," "energy efficiency," and "contributions to customer health and well-being." Furthermore, marketing, sales, and support departments that interact with customers, along with development departments that create customer value (products/services), can boost purchasing intent by promoting initiatives related to "energy," "water," "CO2," and "waste," and communicating the results to customers.

Q. Presenting these analysis results with concrete data is a new approach for the industry, isn't it?

Sakai: This is indeed the first time ISID has publicly announced "ESG that leads to tangible results." While one application is certainly to engage in ESG initiatives aligned with the "demanded items" revealed by this analysis, we hope even more that visualizing, through numbers, how "ESG initiatives bring positive benefits to the company" will encourage corporate adoption of ESG. We want to fully support the motivation that arises from this.

By preparing the external environment, more convenient and user-friendly data analysis becomes possible

Q. Could you share your future outlook? For example, will it be possible to address requests like "We want to simulate the effects of this ESG initiative on specific segments"?

Kanie: Yes, that's correct. As we collect more data and machine learning advances, such simulations will become feasible. We also aim to further enhance the value we provide to client companies. This means not only suggesting themes with expected impact but also actively supporting the concrete implementation of those themes.

Sakai: We'll start with the basics. Earlier, Matsuyama mentioned in the analysis summary that Japanese companies haven't established the same level of commitment to information disclosure as their overseas counterparts. As disclosure improves, the volume of analyzable data will increase, which should naturally enhance the precision of our data analysis.

To that end, we also want to research the potential impact of analyzing and publishing similar ESG initiatives from overseas companies. Ranking ESG value is another intriguing area. Analyzing the characteristics of companies with high ESG value rankings or strong growth potential, or compiling the traits of growing companies and verifying their correlation with ESG initiatives, would be fascinating endeavors.

As Mr. Kanie mentioned earlier, if we can enhance our data analysis techniques—such as improving machine learning accuracy—to the point where we can simulate "how much performance can be expected in the medium to long term from this initiative," it could bring significant change to the nature of ESG management in Japan.

Matsuyama: Going forward, I hope we can not only analyze data but also use the knowledge gained from analysis to deepen our insights. For example, there are gaps in results between analyses targeting overseas companies and those targeting Japanese companies. It would be valuable to properly examine why this occurs. Doing so should help clarify what Japanese companies need to do regarding ESG.

We've seen that combining CALC with machine learning for ESG management data analysis can reveal effective ESG initiatives tailored to specific targets, concretely presenting the ESG actions companies should pursue. Visualizing such data should clarify what companies need to prioritize now and what information they should communicate, enabling them to implement more effective measures. As data analysis advances, we can expect Japanese companies' ESG management to develop further.

※CALC is a registered trademark of Sony Group Corporation.
※CALC is a technology developed by Sony Computer Science Laboratories, Inc.