How results are calculated
The calculation is being documented against the platform and the results workbook, and a few of the figures below are still being confirmed with our product team. Where this page and the numbers in your Excel export disagree, the export is the record. Tell us through the in-app chat and we correct the page.
Know exactly how a score becomes a materiality conclusion, so you can defend your results.
Introduction
The calculation has six steps. Each one is reproduced in the Excel export that comes with your results, sheet by sheet, so an auditor can follow a number from a single answer to the final matrix.
A funnel diagram: answers, stakeholder group scores, IRO scores, subtopic scores, topic scores, thresholds.
Step 1: Score per answer
Every answer to a question is turned into a score between 0.2 and 5:
| IRO type | Score |
|---|---|
| Positive impact | Average of scale and scope, multiplied by likelihood |
| Negative impact | Average of scale, scope and irremediability, multiplied by likelihood |
| Financial risk or opportunity | Magnitude multiplied by likelihood |
Likelihood enters as a percentage (20%, 40%, 60%, 80%, 100%). A respondent who gave an overall score instead of a breakdown has that score used directly. See How to score IROs.
Skipped questions and "I don't know" answers are excluded. They do not count as 0.
Step 2: Weight per answer
Each answer carries a weight:
- Expertise of the respondent: 1 (little or no knowledge), 2 (stays informed), 3 (expert). For a focus group, the number of participants.
- Method: survey 1, interview 3, focus group 6.
Answer weight = expertise × method. A focus group of seven participants has a weight of 42, a survey respondent with little knowledge a weight of 1.
Step 3: One score per stakeholder group and IRO
Within each stakeholder group, the answers to one IRO are averaged, weighted by the answer weight. This gives one representative score per group per IRO. The method weight matters here: when a group answered through both a survey and interviews, the interviews count more. When a group used a single method, the method weight has no effect.
Step 4: One score per IRO
The group scores for one IRO are combined, weighted by the topic weight you set per group (Low 1, Normal 3, High 6):
IRO score = sum over groups of (group score × topic weight) ÷ sum of topic weights.
This is where one group counts more than another. A group that only answered a survey and a group you interviewed in depth count the same unless you set different topic weights.
Step 5: Subtopic and topic scores
Per subtopic and per IRO type, the subtopic score is the highest IRO score. Positive impacts are never offset against negative impacts; opportunities are never offset against risks.
The subtopic scores are then rescaled to 1 to 5 with a min-max normalization, separately for impact and for financial materiality. The lowest subtopic score on an axis becomes 1 and the highest becomes 5. This is the only normalization in the method: scores stay on their raw scale through the stakeholder and IRO steps above. It corrects for respondents who score everything high, or everything low, and makes the two axes comparable. It is done across the ESRS subtopics only; custom subtopics are shown but do not shift the scale.
Impact materiality of a subtopic = the higher of its positive and negative impact scores. Financial materiality = the higher of its risk and opportunity scores. A topic takes the highest score of its subtopics on each axis.
Because the 1 to 5 scale is anchored on subtopic scores, an individual IRO rescaled onto it can fall outside the scale. Such IROs are shown at 1 or 5. IRO-level scores are indicative; subtopic and topic scores are the reported figures.
Step 6: Thresholds and conclusions
The materiality threshold is calculated per axis:
- Take the average of the topic scores on that axis per IRO type. For impact materiality, each topic contributes its highest positive impact score and its highest negative impact score as two separate values; for financial materiality, its highest risk score and its highest opportunity score. Positive and negative, and risk and opportunity, are not first combined into one score per topic. Custom, entity-specific topics count here, even though they are left out of the normalization pool in step 5.
- If that average is below 3, the threshold is 3. The calculated average is never used below the midpoint of the scale: EFRAG treats the midpoint as the lowest reasonable bar for calling a topic material, so a low-scoring assessment does not end up with a low bar that makes everything material. For assessments where stakeholders scored conservatively, this floor is the rule that actually applies on one or both axes.
The comparison is made on the normalized score, the 1 to 5 score from step 5, not the raw score. A subtopic or topic whose normalized score is at or above the threshold is material on that axis; exactly at the threshold counts as material. Below it, it is not. A subtopic or topic is material overall when it is material on at least one axis.
Below the threshold sits an undecided band: from the threshold minus a margin, 0.5 by default, up to the threshold. A result in the band is flagged for your decision; see Decide on undecided topics. With a margin of 0 there is no undecided band: a result is either material or not material.
Both thresholds and the margin are shown on the results page and written into your DMA report and the Materiality summary sheet of the Excel export, so you can always see which of the two rules, the average or the floor of 3, produced the number you are looking at.
The platform's conclusion at every level, IRO, subtopic and topic, follows this rule. Nothing is inherited from a higher level.
What is not in the calculation
- AI scores. The AI pre-assessment and the AI IRO scores are never an input to your results. They are shown next to the results for comparison.
- The review of your own reporting. The evidence drawn from the documents you uploaded is a second evidence base in the DMA report, never an input to a score, a threshold or a conclusion.
- Custom topics. IROs on entity-specific topics are scored and shown, but they are not part of the ESRS normalization pool.
- Population size of a stakeholder group. It is informative only.
Changing the parameters
Thresholds, the undecided margin and the method weights are set by Karomia. Ask us through the chat if you want to adjust them, for example to set both thresholds at 3, or to remove the undecided band. Every change is written in your DMA report. Topic weights per stakeholder group you change yourself; see Review your results.
Output
- The ability to trace any number in your matrix back to a single stakeholder answer, which is what an auditor re-performs.
- The vocabulary for the parameters printed in your DMA report: weights, normalization, thresholds and the undecided margin.