Most organisations approaching CSRD for the first time ask the same question: what do we have to report? It is the wrong first question. The answer depends entirely on your materiality assessment, which means the assessment is not a step in the process — it is the step that defines the process.

Two lenses, not one

Double materiality means assessing two things about every topic, and they are genuinely different questions:

  • Impact materiality — how your operations and value chain affect people and the environment, whether or not that affects your finances
  • Financial materiality — how sustainability matters affect your financial position, performance and cash flows

A topic is material if it passes either test. That catches people out. Water stress in a supply-chain region may have no measurable effect on your accounts this year and still be material on impact grounds. Conversely, a carbon price you will face in three years is financially material long before it appears anywhere in your operations.

A sequence that works

The regulation describes what to produce, not how to get there. This is the order I use:

  • Map the value chain first — upstream, own operations, downstream. You cannot assess impacts in a chain you have not drawn.
  • Build a long list of topics from ESRS, sector guidance and your own risk register. Start broad; narrowing is the next step, not this one.
  • Engage stakeholders properly. Affected communities, workers and suppliers are not a box to tick — their input is evidence that the assessment considered impacts you would not see from head office.
  • Score each topic on both lenses, using thresholds you have written down before you start scoring.
  • Document the process itself. The assessment output matters less to an assurance provider than being able to show how it was reached.

The three failures I see most

Treating it as a survey exercise. Sending a questionnaire to a stakeholder list and ranking the averages is not a double materiality assessment. It captures perception, not impact, and it will not survive assurance.

Skipping the value chain. Most material impacts for a manufacturer sit upstream. An assessment scoped to your own sites will systematically miss them, and the gap is obvious to anyone reading it.

No audit trail. The single most common finding. The conclusions may be entirely reasonable, but if you cannot show the inputs, the scoring and who decided, the assessment cannot be relied upon.

What good looks like

A defensible assessment produces a documented topic list with a stated rationale for each inclusion and — just as important — each exclusion. Excluding a topic is a legitimate outcome. Excluding it without recording why is not.