Making Sustainability Data Work Harder
How would you describe the current state of sustainability data and reporting within listed companies?
Sustainability has found itself under new management. Most sustainability teams at the listed companies I work with now sit within, or are moving towards, the office of the CFO.
The ESG controller role originally emerged as a bridge between sustainability and finance. Increasingly, however, companies have a reporting manager or someone in a similar role reporting to a finance director under the oversight of the CFO. This change has largely been driven by regulation and the growing requirements around assurance and auditability.
With that shift has come much greater scrutiny of the data itself. CFOs are now having to ask questions about matters such as the gender composition of teams or electricity bills from factories on the other side of the world, so that these can be converted into tonnes of carbon dioxide equivalent.
The contrast with financial data is striking. Financial information benefits from established structures, a long history of governance and a relatively stable technology stack. Sustainability data often comes from multiple new sources, varies considerably in quality and may be incomplete, requiring estimates to fill the gaps. This is not what finance functions have traditionally been accustomed to managing.
Companies are therefore building new processes, supported by technology and, increasingly, AI. But this remains a new frontier.
At the same time, the data has never been more valuable or more necessary. After several years of regulatory uncertainty, the landscape is beginning to stabilise around two broad camps: ISSB-aligned and CSRD-aligned requirements. Companies now have a firmer understanding of what they will need to report and the likely timelines involved.
In the age of AI, this information also has significant strategic value. Companies are sitting on a goldmine of proprietary data which, if managed and surfaced effectively, can provide a genuine competitive advantage. That is the area on which I have been working most closely.
Where can AI make the biggest practical difference to the way companies collect, process and report sustainability information?
The obvious answer is report writing and backward-looking analysis, but AI capabilities in those areas are fast becoming commoditised. Before long, almost every reporting platform will have a chatbot capable of producing a draft against whichever framework a company uses.
The more significant opportunity today is in data collection.
Sustainability teams still spend a considerable amount of time reading electricity bills, supplier spreadsheets and invoices, and then manually entering the numbers into a system. AI can already do much of that work. It can extract figures from documents, compare them with previous years and flag gaps that may require an estimate.
Data is often classified as missing when it is not truly absent. It is simply hidden in formats that nobody has had the time or capacity to process.
The next step will be the use of agents that request the data themselves, following up with suppliers and operating sites at a scale that a small central team could never achieve. I am beginning to see this being tested.
As these capabilities become more widely adopted, sustainability data should arrive more quickly and be more complete. Sustainability management can then move beyond a once-a-year, backward-looking exercise and begin to inform decisions as they are made, including the pricing of physical and transition risks.
Are investors primarily asking for more sustainability information, or for a smaller number of more reliable, comparable and decision-useful measures?
Investors increasingly want a smaller set of measures that are more decision-useful.
There is now broad recognition that a company’s valuation can be linked directly to its sustainability performance in several ways. This has led to greater interest among investors in understanding how sustainability-related data can inform investment decisions.
Much of the available information remains backward-looking. However, with the climate changing rapidly and many of the consequences still uncertain, historic data and trend lines will only take investors so far.
As sustainability data becomes more current, accurate and trusted, it will begin to show where a company is heading, rather than simply where it has been. That should help investors identify the relatively small number of metrics that genuinely matter for each type of business, rather than asking every company to disclose everything.
Transition plans will become central to this assessment. As the scale of change required to address climate risk becomes clearer, a gap is likely to emerge between companies that are actively managing that risk through a credible transition plan and those that continue to leave it unaddressed and unpriced.
Investors are likely to be the first to penalise companies they regard as unprepared for what lies ahead.
How are company scale and resources affecting the deployment of AI in preparing and analysing sustainability data?
Across my client base, I see three broad patterns of adoption, largely reflecting company size:
Smaller companies
In smaller businesses, AI adoption is often driven by a single individual who is using the available tools to solve problems one at a time. That person may be trying to stretch a small team’s capacity across everything from bookkeeping to invoicing and reporting.
The immediate objective is typically process improvement and efficiency. Smaller companies also rely heavily on software designed for their part of the market, such as platforms that calculate a business’s carbon footprint. Those providers are now embedding AI capabilities into their products and passing them on to users.
The result is a combination of employees using AI individually and AI arriving through the software the company already has.
Mid-sized companies
When they get it right, mid-sized companies can use their agility to remain at the forefront of AI adoption.
However, alongside approved corporate tools, I often see employees using alternative tools that they prefer, sometimes through personal accounts. Sustainability information can be highly sensitive, so this creates a growing security and governance risk.
The challenge for mid-sized companies is to preserve their ability to experiment and switch to better tools as the market develops, while putting appropriate governance around their use.
Large enterprises
At enterprise level, the prevailing mood is more hesitant. Many senior leaders continue to see AI primarily as a risk, with returns that have yet to be proven. Recent headlines about companies seeing limited benefits from significant AI expenditure have reinforced that concern.
Part of the problem is the way AI has been introduced. Software providers in finance and sustainability have adopted different approaches, but two stand out.
The first is to build agents around specific workflows rather than simply adding a chatbot to an existing platform. This makes the operational improvement measurable, helps demonstrate the return on investment and makes the agents easier to train.
The second is to establish the governance before deploying the technology at scale, even if that means moving more slowly than competitors. This can help earn the CFO’s confidence that AI can be used without placing sensitive data at risk.
That confidence matters because technology leaders are understandably concerned about the implications for their company’s intellectual property. Several have told me privately that they worry connecting a frontier model to all the information that would make it genuinely useful could, in effect, create a pipeline through which the model learns their business. They question whether that could ultimately weaken the company’s control over its own knowledge and capabilities.
As investors increasingly use their own AI tools to interrogate reports, what does that mean for how companies prepare and publish sustainability information?
Companies need to recognise that a human may no longer be the first audience for their reporting. Before long, the human may not even be the most influential audience in the initial stages of the information journey.
By the time an analyst receives a morning briefing, the underlying information may have passed through several layers of AI interpretation. It could have been collected by a research agent, processed by a data provider, and then summarised by a personal tool configured for that particular analyst.
As a result, understanding what information ultimately surfaces is becoming much more difficult.
During the development of search, this became relatively straightforward. Companies came to understand, at least broadly, how search algorithms worked and what practices improved or reduced visibility. We are now entering a period in which the criteria for visibility are shifting again, while companies still need to ensure that their information reaches the right people at the right time.
Companies will therefore need to adapt how they present and publish information as quickly as possible while these new conventions develop. What does ChatGPT surface first, compared with Claude? Does Grok interpret a particular format differently from Gemini? These are becoming highly relevant questions for IR teams seeking to maximise the visibility and accessibility of their disclosures.
The human audience nevertheless remains essential. Behind these systems, people will still be making judgements and investment decisions.
I expect this to lead to multiple versions of the same underlying document: an agent-to-agent version and a human-to-human version. We are already seeing interesting examples of AI agents communicating with one another in novel ways. A company’s ability to understand and harness these new forms of communication will become increasingly important when preparing information for investors.
What do you think a genuinely AI-enabled, decision-useful sustainability reporting model will look like?
I envisage a lean, highly technology-enabled team supported by multiple AI agents. Those agents will be able to operate across the company and its supply chain, collecting the data required for both reporting and decision-making.
They will operate under strict protocols governing data visibility, collection and processing. Governance will increasingly sit at agent and system level, rather than depending solely on the actions of individual users.
Teams will be able to create dashboards easily, surfacing the information required for different decisions. The data will feed directly into pricing and investment mechanisms within the finance function.
Ultimately, every significant decision could incorporate a risk-adjusted carbon price. This would give decision-makers a fuller and more accurate picture of the implications of the choices they are making, whether in relation to strategy, mergers and acquisitions, capital expenditure or supplier selection.
The agents would deposit the information into a single, trusted source of truth, such as a governed data lake, where processing and validation take place. Reports would then become a by-product of that underlying data system rather than an isolated annual exercise.
Companies would integrate the reporting frameworks they follow, together with their own policies and methodologies, into the systems used by the models that draft the reports. People would continue to review, challenge and approve those drafts through the appropriate governance chain.
Reporting would become significantly leaner, with different versions of the same underlying information produced for whichever audience requires it, whether human or machine.
What practical steps should an IRO take now to prepare for the growing use of AI in sustainability data and reporting?
IROs should accept that there is a new reader between the company and its investors, and start working with it. There are four practical steps they can take now.
1. Read your reports as an investor’s AI would
Ask two or three leading AI models to summarise your sustainability story and answer the ten questions investors ask most frequently.
Where the models get something wrong, fix the source rather than simply correcting the summary. Make sure each important number has one clear definition and is presented consistently across the annual report, sustainability report and corporate website.
Repeat the exercise before each results season.
2. Make the key data machine-readable
Publish your most important sustainability metrics in a format that machines can interpret easily.
A simple table on the IR website may be more useful than a chart embedded in a PDF. Each figure should be accompanied by the relevant year, reporting boundary and methodology.
IROs should test different formats and develop an evidence-based understanding of which approaches currently perform best.
3. Understand the provenance of your most important numbers
Identify the five sustainability figures most frequently quoted by investors and analysts.
For each one, establish:
- who owns it;
- how it is calculated;
- what it includes and excludes;
- which methodology is used; and
- how quickly it could be produced if an investor requested it today.
An approved AI tool can use existing methodology notes to draft a one-page data card for each metric, which the relevant data owner can then check.
IROs should use these cards to gain access to the discussions in which sustainability data is governed and controlled, rather than participating only at the point when it is presented externally.
4. Use AI in investor preparation
Within the company’s approved technology environment, give an AI tool the previous year’s investor questions alongside the company’s published reports. Ask it to draft responses to the sustainability questions most likely to arise.
Then measure the time saved. Evidence such as, “This process previously took 100 hours and now takes 50,” is precisely the kind of tangible outcome that investors and boards will want to see.
Above all, IROs should treat sustainability data as what it is becoming: the most underused chapter of the company’s equity story.
About Mike
Mike Bascombe is a senior leader within the sustainability community, having worked both in-house and on the consulting side. He focuses on where AI, SaaS, non-financial data, and operational reality meet. He leads sustainability at VantagePoint, where his team builds new operating models and delivers technology directly to listed and private markets clients, making their operations more efficient, reliable and resilient. He has previously worked across the strategy side of sustainability as well as reporting, counting Google, Meta, Amazon, Paramount, and Elastic as clients.