Running a Double Materiality Assessment: Judgement, Method & AI

Programme information

A double materiality assessment (DMA) shows how the company materially impacts nature and people on the one hand, and how the company is impacted economically by nature and people on the other. The DMA is the fundamental ground work that needs to be performed to understand, what needs to be done and what to be aware of.

A DMA has done its job when leadership can act on it, when it hands them clear decision data for prioritising strategic initiatives, and those priorities connect to the strategy the company is actually pursuing. Getting there takes more than a topic list. The assessment has to be anchored in the organisation rather than produced alongside it. It has to rest on a genuine understanding of the whole value chain and the impacts, risks and opportunities inside it, and on an equally genuine understanding of the stakeholders, what they prioritise and how much influence they carry. Its scoring system has to stand up over time and under external criticism. And it should be validated by experts outside the process. No template delivers any of that.

This 9-hour masterclass is built for the person doing the work. It follows the full arc of an assessment, i.e. mandate, value chain and stakeholders, the long list of impacts, risks and opportunities, impact scoring, financial scoring, threshold and prioritisation, report and audit trail and concentrates on the decisions that are genuinely difficult: how to set a scoring key that produces the same answer next year, what evidence stands up when the auditor disagrees, how to handle a value chain that will not give you data, and what to do when leadership declines to prioritise something your analysis says is material.

The regulatory ground has shifted. Directive (EU) 2026/470 narrowed the scope of the CSRD, removed sector-specific standards and reasonable assurance, and the revised ESRS adopted in July 2026 sharply reduced the number of mandatory datapoints. Double materiality itself survived all of it, and it remains the mechanism that determines everything an undertaking reports. At the same time, a growing group of companies now runs an assessment for reasons that have nothing to do with a legal obligation, like customer demands, risks of supply.  The companies that already made the DMA stated that it was what actually transformed the business and its priorities.

Artificial intelligence runs through the programme as a working tool rather than a topic. Each module identifies where a model genuinely accelerates the work, like generating a value chain hypothesis when no one internally can describe it, drafting scoring keys, producing a challenge-set of industry impacts, analysing free-text stakeholder responses, assembling documentation and references. The closing module covers how to document AI use so that it survives review by an auditor.

The objective is not familiarity with the concept of double materiality. It is to leave able to run an assessment end to end, to defend the judgements inside it, and to recognise the specific points where a specialist, like a climate scientist, a human rights practitioner, a finance lead, a lawyer needs to be brought in rather than approximated.

Structure

Module 1: Mandate, Scope and the Foundation for the Analysis

Establish the basis before any analysis begins: who the sponsor is, what the assessment is actually for, and what the auditor expects. Cover legal entity structure and capital interests, business model, geography, products and services, and how far the assessment reaches into the value chain. Examine how to plan around key people who have no time, how to ask for information that already exists rather than commissioning new material, and how to get the foundation formally accepted so it cannot be reopened in month three. Includes the proportionality question: what changes between a consultancy, a manufacturer and an importer, and between a company in mandatory scope and one assessing voluntarily.

Module 2: Value Chain and Stakeholders - Where the Findings Actually Come From

The value chain is the single largest source of unknown impacts, and the place most assessments are weakest. Work through how to investigate a value chain rather than receive a document about it: process-level detail, locations, materials, tiers beyond the direct supplier, and what to do when a supplier has no data or will not share it. Then the stakeholder side: choosing between structured questionnaires and 1:1 interviews, anonymity, response rates, and the translation problem, like how a stakeholder's plain-language concern becomes an ESRS sub-topic without the analyst quietly deciding the answer. AI in practice: generating a value chain hypothesis and location-specific impacts where internal insight is missing, and analysing free-text stakeholder responses, with the limits of both.

Module 3: Building the Long List of Impacts, Risks and Opportunities

Assemble a complete and defensible long list. Work through the ESRS topic architecture and its sub- and sub-sub-topics, entity-specific matters that no standard names, the distinction between actual and potential impacts, positive and negative, and the correct placement across upstream, own operations and downstream. Examine the connection between impact identification and human rights salience, and why the sequence “value chain first, industry benchmarks second” matters for completeness. AI in practice: using generated industry impacts as a challenge-set against your own list rather than as a starting point, and testing whether the categorisation you have inherited is actually correct.

Module 4: Impact Materiality - Scoring Keys and Defensible Judgement

The scoring key, not the score, is what an auditor examines. Build keys for scale, scope and irremediability, and the likelihood dimension for potential impacts, using concrete thresholds, like percentages, numbers, defined conditions, rather than adjectives. Examine the weighting of severity for human rights impacts, when a company-specific reference frame is right and when a global one is, how to keep scoring consistent year on year, and how to document assumptions and uncertainty so the analysis is reproducible by someone else. AI in practice: generating and qualifying scoring keys, structured scoring with clarifying questions, and how to review a machine-produced justification for relevance, correctness and completeness.

Module 5: Financial Materiality - Risks, Opportunities and the Finance Conversation

The weakest section of most assessments. Cover how impacts and dependencies convert into risks and opportunities, how to size magnitude and likelihood across short, medium and long horizons, and how to set financial thresholds that connect sensibly to the thresholds used in financial reporting. Work through the conversation with the finance function on what data exists, what is a legitimate estimate, and what is a placeholder pretending to be a number. Examine why financial scoring is a finance discipline that should not be delegated to a model, and what AI can legitimately contribute at this stage.

Module 6: Threshold, Prioritisation, Reporting and the Audit Trail

Convert analysis into a decision and a defensible record. Cover threshold setting and materiality selection, running a prioritisation workshop with a mixed group rather than presenting a conclusion, and what to do when leadership declines to prioritise something the analysis says is material. Move from the material list to the datapoints that follow from it under the revised ESRS, and through the report structure: management summary, material topics, risks, opportunities, recommendations, and the appendices that carry the evidence. Closes with the governance of AI use necessity, model versus expert, verification, provenance, and how to document it for the auditor  and with the recognition question: identifying the specific points in your own assessment where you need help rather than a better prompt.

‍