ESG Reporting: Best Practices for Modern Businesses

The single most useful thing we tell a client preparing for mandatory sustainability reporting is that they already know how to do this. They run a financial close every quarter. They have a controls framework, an internal audit function, a materiality judgment process and an external auditor who asks awkward questions. Sustainability reporting is that, applied to a different data set.

The companies that struggle are the ones that keep the two worlds apart — where emissions data lives in a spreadsheet owned by one person in facilities, and the report is drafted by a communications team who receive the number as a fact rather than as an output of a process.

Start from the data, not the framework

There is a strong temptation to begin with a standard: read the requirements, build a gap list, work backward. It produces a tidy project plan and a report that is technically compliant and operationally useless.

A better starting point is to inventory what you actually measure today, where it comes from, who touches it, and how confident you would be if someone asked you to prove it. That exercise typically reveals three categories: metrics with a real system behind them, metrics reconstructed annually from invoices, and metrics that are estimates dressed as measurements.

Only the first category can carry a target. The second can, once the reconstruction is documented and repeatable. The third needs a measurement plan before it needs a disclosure.

Build one data set, map it many ways

Companies that maintain separate workflows for each framework end up publishing numbers that contradict each other — a figure in the annual report that does not match the one in a customer questionnaire, which does not match the CDP response.

The fix is architectural. Define each metric once, with a single owner and a single definition, and hold it in one place. Then maintain mappings out to each disclosure destination. When a standard changes, you amend a mapping. When an auditor asks where a number came from, there is one answer.

The most common finding in a first assurance engagement is not an incorrect number. It is two correct numbers that disagree because they were calculated for different audiences.

Controls, before the auditor asks

Limited assurance is not a light-touch review. The assurance provider is testing whether your process could have produced a materially wrong figure, and they will sample evidence to find out. The controls that make that comfortable are unremarkable:

• A named preparer and a named reviewer for every material metric, with the review evidenced
• Source documents retained in a form that can be sampled, not summarized into a spreadsheet and discarded
• Estimation methodologies written down, with the uncertainty stated
• A change log for factors, boundaries and methodologies, with reasons
• A defined recalculation policy that says what triggers a restatement of the base year

None of this is expensive. All of it is difficult to retrofit three weeks before a deadline.

Write about variance, not about ambition

The narrative section is where most reports lose credibility. A chart showing progress against a target, with no explanation of the year the line went the wrong way, invites the reader to assume the worst — and increasingly invites a regulator to ask.

Good disclosure explains variance. Emissions rose because an acquisition added two manufacturing sites; here is the like-for-like figure. Water intensity worsened because a drought forced a change in cooling regime; here is what we are doing about it. That kind of writing is harder and it is the only kind that builds trust with an analyst who has read forty of these.

It also has a practical benefit: a company that explains its variances internally every quarter is a company that notices problems in year two rather than year six.

Connect it to the financial statements

The direction of travel in every major regime is toward connectivity — sustainability information that is consistent with, and cross-referenced to, the financial statements. If your climate transition plan implies retiring an asset early, that assumption should be visible in the impairment testing. If a physical risk is material enough to disclose, it is material enough for the going concern discussion.

Teams that report into finance rather than alongside it find this straightforward. Teams that do not tend to discover the inconsistency when an auditor points it out.

A reasonable sequence

For a company at the start of this, the order that works is: fix the boundary, inventory the data, build the controls, run a dry close, then worry about the framework mapping and the design of the document. Most organizations do it in exactly the reverse order, which is why the first year is always the worst one.