Two inspectors walk into the same factory on the same morning. The first works for the insurer: she is looking for what could hurt the business — the flood plain, the ageing boiler, the single supplier who makes the one component nobody else makes. The second works for the public health authority: he is looking for what the business does to everyone else — what goes into the river, what the night shift breathes, what the trucks do to the road through the village. Same building. Same day. Two clipboards, and two entirely different lists.
Double materiality is the requirement to carry both clipboards. Under the European Sustainability Reporting Standards, double materiality requires an undertaking to assess sustainability topics from two directions — how they affect the company financially, and how the company affects people and the environment — and a topic that scores on either side is material. Not the average of the two. Either.
Neither perspective outranks the other. A methodology that can return “medium impact, medium financial, therefore not material” is implementing a weighted average, not double materiality.
That single rule contains the mistake that turns up most often in real filings, and we will come back to it with numbers from a regulator that read ninety-one of them. First, the ground rules: this guide covers the double materiality method. If you need the framework itself — the twelve standards and how they fit together, who must report, what changed in the 2026 rewrite — that lives in our companion guide to what ESRS actually is.
What the first reporting cycle actually produced, in three numbers:
- 91 — sustainability statements read by national enforcers across 23 Member States, all of them under assurance
- 61.7% — the share that adequately explained how they ran the assessment
- 100% — the share for which climate came out material. Every single one.
⚠️ Regulatory status — verified 23 August 2026. Double materiality itself is unchanged by the 2026 simplification, but the standards around it have moved. The revised ESRS were adopted on 3 July 2026 and are still within the scrutiny period before the European Parliament and the Council. They are adopted but not yet in force; until they clear, ESRS Set 1 as adopted in 2023 remains the legally operative text. Where the two versions differ on the assessment, this guide says which is which.
Quick answer. Double materiality is the ESRS rule that a sustainability topic must be reported if it is material from either of two perspectives — the effect the topic has on the undertaking’s financial position, performance, cash flows or cost of capital, or the effect the undertaking has on people and the environment. Either perspective is sufficient on its own, and neither outranks the other. Impacts are scored on severity and likelihood, gross of the company’s own mitigation; financial effects on magnitude and likelihood. The assessment decides which of the ten topical ESRS standards apply, and it is disclosed through ESRS 2 IRO-1, IRO-2 and SBM-3. The 2026 revision left the principle untouched while permitting a faster top-down route through it.
Who should read this
- Sustainability reporting managers
- Group financial controllers
- Heads of risk and ERM
- Internal audit
- ESG consultants
- Procurement and IT evaluating platforms
- Voluntary reporters facing customer questionnaires
- Anyone preparing for limited assurance
🔑 Key takeaways
- Double materiality means either lens is enough — never average them. A topic is material if it passes the impact test or the financial test. Blending both into one score is the error that produces a tidy matrix and an indefensible conclusion.
- The regulator has already told you what it will look for. ESMA reviewed 91 sustainability statements from 23 Member States. Only around 62% met the objective of the disclosure requirement covering the assessment process, and close to 20% disclosed nothing at all about their thresholds.
- Double materiality assesses impacts gross, before your own mitigation. ESMA could confirm gross assessment for 72.5% of the sample. For the rest it was unclear, or the company had quietly scored net — which flatters every conclusion.
- Non-material topics require disclosure too. Only 54.3% gave sufficient information on topics they concluded were not material, and more than 27% gave none. Concluding “not material” is a disclosure event, not an exit.
- The 2026 revision made double materiality lighter to document and heavier to defend. A top-down route is now permitted, exhaustive IRO coverage is not required, and the prescribed topic list became optional — which shifts the burden of justifying your own topic universe squarely onto you.
- Climate came out material for 100% of the sample. Every one of the 91 issuers disclosed at least one climate-related impact, risk or opportunity. Plan on E1 applying to you.
On this page
| Question | Answer |
|---|---|
| What it is | The test that decides which sustainability topics an undertaking must report on under the CSRD |
| The rule | A topic is material if it passes the impact test or the financial test — either is sufficient |
| Impact lens | Severity (scale, scope, irremediability) combined with likelihood — assessed gross |
| Financial lens | Magnitude of the anticipated financial effect combined with likelihood |
| Legal basis | ESRS 1, under Commission Delegated Regulation (EU) 2023/2772 |
| Worked method | EFRAG IG 1 — four steps, non-authoritative |
| Disclosed through | ESRS 2 IRO-1 (process), IRO-2 (list), SBM-3 (the material IROs) |
| Thresholds | Not prescribed — you define them, and you disclose them |
| Changed in 2026 | Top-down route permitted, exhaustive coverage not required, topic list now optional |
| Unchanged in 2026 | The principle, the two lenses, the four steps |
What Double Materiality Actually Asks
Double materiality predates the CSRD — the European Commission first formalised it in its 2019 guidelines on reporting climate-related information, which stated in terms that the reporting regime then in force carried a double materiality perspective — but the directive is what turned it from a good idea into a legal test.
Under ESRS, the double materiality assessment does two jobs at once, and it is worth separating them because teams routinely conflate the two.
The first job is to decide which of the ten topical standards apply to you. That is the gate: a topic that is not material is a topic you do not report against. The second job is to decide which individual pieces of information are material within the topics that passed — the standards call this materiality at information level, and it is where most of the volume reduction actually happens.
| Level | Double materiality question | Consequence of the answer |
|---|---|---|
| Topic level | Is this sustainability topic material, by impact or by financial effect? | Determines which topical standards you apply at all |
| Information level | Within a material topic, is this specific datapoint necessary to understand the matter? | Determines the length and content of the statement |
The Two Lenses of Double Materiality, and the Line Between Them
The first lens of double materiality looks outward. Impact materiality covers what you do to the world. It captures the actual and potential effects — positive and negative, short, medium and long term — that the undertaking has on people and the environment, across its own operations and its upstream and downstream value chain. Crucially, it applies whether or not those effects ever rebound on the company financially. A factory that pollutes a river cheaply and lawfully has a material impact even if the pollution never costs it a euro.
The second lens looks inward. Financial materiality covers what the world does to you. It captures sustainability matters that generate risks or opportunities with a material influence on the undertaking’s development, financial position, performance, cash flows, access to finance or cost of capital. This is the lens that overlaps with enterprise risk management, and the one your CFO will recognise.
| Impact materiality | Financial materiality | |
|---|---|---|
| Question | What do we do to people and the environment? | What could sustainability do to our numbers? |
| Criteria | Severity — scale, scope and irremediability — combined with likelihood | Magnitude of the anticipated financial effect, combined with likelihood |
| Special rule | For potential negative human rights impacts, severity takes precedence over likelihood | None — magnitude and likelihood are weighed together |
| Assessed | Gross, before your mitigation measures | Considering the risk as it stands |
| Actual impacts | Severity alone — likelihood does not apply to something already happening | — |
| Overlaps with | Human rights due diligence, GRI reporting | Enterprise risk management, IFRS S1/S2 |
The gross rule is where quiet failures start. Impacts are assessed before the mitigation, controls and prevention measures you have put in place — otherwise a company that manages a risk well would report fewer material topics than one that ignores it, which inverts the entire point. ESMA could confirm gross assessment for 72.5% of the issuers it reviewed. For the remainder it was either impossible to tell, or the company had explicitly folded its own controls into the score. If your scoring workshop asks “how bad is this, given what we already do about it”, you are scoring net.
One further distinction is worth holding onto, because it governs what you can be challenged on. Impact materiality does not require a financial consequence; financial materiality does not require an impact. But the two connect more often than teams expect — an impact you cause today frequently becomes a risk you carry tomorrow, through regulation, litigation or reputation. ESRS asks you to consider that connection explicitly rather than to treat the lenses as sealed compartments.
What a Regulator Found in 91 Real Double Materiality Assessments
Most guides to this subject describe the process. Very few describe what happened when companies ran it. In October 2025, ESMA published the results of a fact-finding exercise on materiality disclosures, conducted with national enforcers across the first full cycle of ESRS reporting: 91 issuers, 23 Member States, roughly a third of them applying the standards voluntarily, all of them subject to assurance. Ninety per cent had more than 1,000 employees, and the median issuer employed close to 10,000 people — which makes the sample a reasonable proxy for the population that remains in scope after Omnibus I.
The findings are the closest thing available to an answer key for double materiality. This is what national enforcers concluded when they read the disclosures on the assessment process itself.
| What was tested | Result | What it tells you |
|---|---|---|
| Met the objective of IRO-1 overall | 61.7% | Nearly four in ten did not adequately explain how they ran the assessment |
| Explained their thresholds | close to 80% | 17 issuers — close to a fifth — disclosed nothing about thresholds at all |
| Disclosed input parameters | 70.4% | Assumptions were the least-reported of data sources, scope and assumptions |
| Assessed impacts gross | 72.5% | For the rest, unclear — or explicitly net |
| Consulted stakeholders | 93.8% | But only 71.6% explained adequately who they were and how they were engaged |
| Used due diligence to inform the assessment | 80.2% | The link to due diligence is now standard practice |
| Separated the impact and financial processes | 78% (derived) | ESMA publishes the complement: 22% described a single undifferentiated process |
| Sufficient disclosure on NON-material topics | 54.3% | Over 27% disclosed nothing — the weakest area in the whole exercise |
| Described their material IROs | ~90% | Description is not the problem; specificity is |
| Targets or an explicit negative statement for every material topic | 71.4% (derived) | 28.6% left at least one material topic with neither a target nor a statement explaining its absence |
Percentages are of the sample reviewed. Rows marked “derived” are complements we calculated from a published figure — the ledger at the end of this guide identifies every one.
Two double materiality results deserve to be pulled out of the table, because they are the ones that turn into audit findings.
The first is the non-material topics gap. Deciding a topic is not material does not release you from writing about it — the standards require information about how that conclusion was reached, and the topical standards carry their own process datapoints. Barely half the sample handled this adequately, and more than a quarter skipped it entirely. It is the single most reliable place to find a deficiency in a first-year statement, and the reason is structural: teams instinctively treat “not material” as the end of the work rather than the start of a disclosure.
The second is boilerplate double materiality prose. ESMA’s recurring complaint was not that companies failed to describe a process, but that they described the process from the guidance rather than the process they actually ran — reproducing EFRAG’s illustrative scoring steps without explaining which entity-specific considerations fed into them. Enforcers singled out as good practice the companies that said what their scale actually meant: for the scope component of severity on an environmental impact, “low” meaning one location, “medium” a few, “high” widespread, “very high” everywhere. That is four extra sentences, and it is the difference between a disclosure that survives review and one that reads as though it could belong to any company in Europe.
If your assessment concludes that climate, own workforce and business conduct are all immaterial, you are not wrong by definition — but you are outside the observed behaviour of every comparable filer, and you should expect to be asked why.
One more finding is worth flagging because it will shape the next cycle. The largest cluster of entity-specific topics — matters companies reported that do not appear anywhere in the ESRS topic list — related to data, cybersecurity and the use of AI, disclosed by 26.4% of the sample. Practice diverged: some treated it as a standalone entity-specific matter, others folded it into consumers and end-users or business conduct. There is no settled answer yet, which means whichever route you take needs an explicit rationale.
The Four-Step Double Materiality Process
ESRS 1 sets the double materiality requirements; EFRAG’s Implementation Guidance 1 describes a worked process. The distinction matters more than it appears: IG 1 is non-authoritative. It illustrates how the standards may be applied without creating new obligations, and where guidance and standards diverge, the standards win. Following IG 1 is not compliance — it is evidence of a defensible method.
| Step | What you produce | Where it fails |
|---|---|---|
| A. Understand the context | Business model and value chain mapped by activity and geography; affected stakeholders identified; users of the statement identified | Value chain mapped to tier 1 and stopped — a limitation enforcers saw flagged in assurance opinions |
| B. Identify IROs | A candidate list of impacts, risks and opportunities, each categorised actual or potential, positive or negative, and located in own operations or value chain | Candidate list generated from the topic list alone, so nothing entity-specific ever surfaces |
| C. Assess and score | Severity and likelihood for impacts; magnitude and likelihood for risks and opportunities; thresholds applied | One blended score; thresholds set after seeing the results |
| D. Report and document | Disclosures under ESRS 2 IRO-1 and IRO-2, material IROs under SBM-3, plus the audit trail behind all of it | The statement describes the guidance, not the assessment that was run |
Two practical notes on step A of the double materiality process. Stakeholder engagement may be direct or, where direct engagement is not feasible, conducted through proxies or experts — the standards do not require you to survey every affected community, but they do require you to say which route you took and why. And for groups, material topics identified at subsidiary level have to find their way into the group assessment; the final version of IG 1 added guidance on exactly this point because multinationals kept asking.
On step D, the disclosure vehicles are fixed and worth naming, because software demos frequently show a beautiful matrix and no path to any of them. IRO-1 describes the process: methodology, thresholds, input parameters, stakeholder involvement, the relationship to due diligence. IRO-2 lists the disclosure requirements complied with and where to find them. SBM-3 describes the material IROs themselves and how they interact with strategy and business model.
Ask the vendor: show me the export that populates IRO-1 paragraph 53 — thresholds, input parameters, stakeholder engagement and due diligence linkage — as disclosure text, not as a chart. If the answer is a screenshot of a matrix, the platform stops one step short of the deliverable.
Setting Double Materiality Thresholds That Survive Assurance
ESRS does not prescribe a double materiality scoring methodology, and it does not set a numeric threshold. That freedom is the most misunderstood feature of the framework: it is not permission to leave thresholds implicit, it is an obligation to define and disclose your own. Most issuers in the ESMA sample built a scale — typically five levels — and set the threshold at the score above which a topic became material.
Enforcers noted the strongest disclosures on the financial side, where companies named the nature of the threshold — a percentage of revenue or of EBITDA — and some published the number itself. On the impact side, disclosure quality dropped, because a severity scale is harder to anchor to anything external.
Here is a sequential self-audit you can run against your own double materiality methodology in under an hour. Stop at the first one you fail — everything after it inherits the defect.
- Were the thresholds written down before the scoring workshop? Check the file dates, not your memory. A threshold set after the scores exist is a rationalisation, and it is the first thing a sceptical reviewer looks for.
- Do impacts and financial effects have separate thresholds? Two lenses, two scales, two cut-offs. One combined threshold cannot implement a rule where either side is sufficient.
- Can you state what each level of your scale means in your own operations? Not “significant impact” — “affects one site”, “affects several sites”, “affects all sites”. If the definition would fit any company in your sector, it will read as boilerplate because it is.
- Were impacts scored before your mitigation? Pull three scored impacts at random and ask the person who scored them whether existing controls influenced the number. The honest answer is often yes.
- For every topic you concluded was not material, is there a written basis? Not a blank cell in a spreadsheet. A sentence, with the input that supported it.
- Does every material topic have a policy, an action, a target and a metric — or an explicit statement that there is none? The negative statement is a disclosure, not an omission. Nearly three in ten filers missed this.
A threshold you cannot defend is worse than a conservative one. The asymmetry runs one way: a threshold set too low produces a longer statement and more work. A threshold set too high, or set after the fact, produces a statement missing a material topic — which is the failure mode that generates qualified opinions. Two of the 91 issuers received a qualified opinion, and one of those pointed at the materiality assessment. Four more drew an emphasis of matter on materiality considerations, including one flagging that the assessment stopped at tier 1 of the supply chain.
What the 2026 Revision Changed Inside the Double Materiality Assessment
The headline from the July 2026 rewrite was volume: mandatory datapoints cut by more than 60%. Inside the double materiality assessment, the changes are smaller in number and larger in consequence, and they run in a consistent direction — less prescribed procedure, more required justification. The full picture of what the 2026 revision changed sits in the companion guide; what follows is only what touches the method.
| Change | What it means in practice | Status |
|---|---|---|
| Top-down route permitted | You may start from business model and strategy to identify clearly material topics, rather than working bottom-up through every topic from scratch. Bottom-up remains valid. | The 2026 differentiator |
| Exhaustive IRO coverage not required | You are not obliged to assess every possible impact, risk and opportunity across all operations and the value chain | Newly critical |
| Reasonable and supportable information | Information available without undue cost or effort is sufficient — a general relief rather than a narrow phase-in | Newly critical |
| Topic list becomes optional | AR 16 moves to Appendix A and its use is no longer mandated; the sub-sub-topic layer is removed and terminology shifts to topics and sub-topics | Emerging |
| Annual reconsideration | Consider each year whether significant changes to activities, structure, relationships, methodology or the external environment would alter your conclusions | Proven |
| Climate non-materiality justification | Retained but scaled back — see the caution below | Underrated |
| Double materiality itself | Unchanged in principle and in structure | Commodity |
On climate, check which version you are planning against. Under Set 1 — still the legally operative text — concluding that climate is not material requires a detailed explanation including a forward-looking analysis. The revised standards delete that provision from ESRS 1 and move a lighter version into ESRS 2: the basis for the conclusion must still be disclosed. Either way, climate is the only topic whose omission must be argued rather than merely recorded, and it was material for 100% of the issuers ESMA reviewed. We set out this correction in full in the companion guide’s section on ESRS E1, including which claim traces to which text.
The through-line in these changes is easy to state and uncomfortable to act on. The revision removed procedure and left judgement. When AR 16 was the mandated starting point, your topic universe was somebody else’s decision and you could point at the regulation. Now the universe is yours to define, which is faster, cheaper — and yours to defend. Companies reading the rewrite as a licence to do less analysis have read the direction of travel backwards.
Where Double Materiality Disappoints
A guide that only explains how double materiality works is marketing. Four things about it are genuinely unsatisfying, and knowing them in advance is what stops a project from stalling in month four.
Double materiality does not produce comparability, and it was never going to. Two companies in the same sector, with similar operations, can reach different material topics because they set different thresholds, engaged different stakeholders and drew their value chain boundary in different places. ESMA said this directly: divergence on gross versus net assessment may hinder comparability between two issuers in the same sector. Investors who expected the standards to deliver like-for-like screening are still waiting, and the 2026 revision — by making the topic list optional — moved slightly further from that goal, not closer.
The double materiality matrix is a communication device, not a method. The four-quadrant chart every consultant produces is not required by the standards, appears nowhere as a disclosure requirement, and hides the two things that actually matter: the threshold line and the reasoning behind each score. Enforcers did not ask to see matrices. They asked what the scale meant and who decided.
Stakeholder engagement is the requirement most likely to become theatre. Nearly 94% of the ESMA sample reported consulting stakeholders. Only 71.6% adequately explained who those stakeholders were, how they were engaged, and how their input fed the double materiality conclusion. The gap between those two numbers is the space where a survey sent to a customer mailing list gets described as engagement with affected communities. The standards allow indirect engagement through proxies and experts precisely because direct engagement is often impractical — which means there is no excuse for describing something you did not do.
Double materiality effort does not scale down with the disclosure volume. This is the structural disappointment of the 2026 rewrite. You will write substantially less. The assessment behind it did not get smaller, and the parts that were removed — prescribed steps, mandated topic lists — were the parts that used to do some of your thinking for you. Budget for the same analytical hours and a shorter document.
A Sensible Double Materiality Adoption Sequence
Sequenced by dependency, not by ambition — this is the order a double materiality project survives. Each double materiality step produces the input the next one needs, which is why reordering them is the most common cause of a re-run.
- Fix the reporting boundary first. Which legal entities, which consolidation basis, how far into the value chain, and what you will do about tier 2 and beyond. Every score depends on this, and changing it later invalidates everything downstream.
- Write the thresholds and the scale definitions. Before any workshop, before any scoring, in a dated document. This is the single highest-return hour in the entire project.
- Decide top-down or bottom-up, and record why. Top-down is faster where topic-level materiality is evident from the business model. Bottom-up remains appropriate where a topic needs deeper analysis. Mixing them is allowed; failing to say which you used is not.
- Build the IRO inventory, then challenge it for what is missing. A list generated from the standard topics will never surface an entity-specific matter. Ask explicitly what this business does that no standard topic describes — the data, cybersecurity and AI cluster surfaced this way for a quarter of the ESMA sample.
- Score impacts gross, separately from financial effects. Two workshops or two passes, never one blended session, and never with the mitigation on the table.
- Document the non-material conclusions as you go. Writing them at the end, from memory, is how the weakest area in the ESMA findings gets created.
- Map every material topic to a policy, action, target and metric — or an explicit negative statement. Do this before drafting, not during, so the gaps are visible while there is still time to close them.
- Only then choose or configure software. A platform bought before steps 1 to 3 will impose its own boundary, scale and threshold logic on you, and you will spend the project arguing with defaults instead of with the business.
On that last point: the tooling question is downstream of the method, not a substitute for it. What a platform needs to do here is narrow and specific — hold the audit trail, version the conclusions, and produce IRO-1 text rather than a picture. We cover the category in CSRD & ESRS reporting, the data layer underneath it in ESG data management, and the scoring approach we apply to every platform in our Evaluation Framework™.
Common Double Materiality Mistakes
Method errors — before anything is written
- Blending the two scores into one, producing a clean matrix and an untraceable conclusion
- Scoring net — letting existing controls reduce the severity of an impact
- Setting thresholds after the scores exist, then having nothing dated to disclose
- Stopping the value chain at tier 1 — specifically flagged in an assurance opinion
- Treating the assessment as a project rather than a standing annual obligation
Disclosure errors — the work was done, the writing lost it
- Describing EFRAG’s guidance instead of the assessment you actually ran
- Going silent on non-material topics — the weakest area in the whole ESMA exercise
- Leaving material topics without targets or a negative statement (28.6% of filers)
- Using entity-specific disclosures where a standard disclosure exists — or never labelling them
- Copying the SBM-2 stakeholder description into IRO-1, which satisfies neither
On scoring net, one number is worth remembering. 27.5% of the ESMA sample — the derived complement of the 72.5% confirmed gross — could not demonstrate they had avoided it. That is more than one in four inventories where an enforcer could not tell whether the company had scored the world as it is or as its own controls leave it.
The Bottom Line
Two inspectors, two clipboards, one building. The framework’s demand is that you carry both and refuse to staple them together — and almost everything that goes wrong in a double materiality assessment can be traced to a team that quietly merged the clipboards because one list was easier to defend to the board than two.
The 2026 revision did not change that demand. It removed some of the scaffolding around it: fewer prescribed steps, an optional topic list, a permitted top-down route, no obligation to assess everything everywhere. What remains is the harder half. You now decide the topic universe, the thresholds, the scale and the boundary — and you disclose all four in a document a national enforcer may read line by line, as ninety-one companies have already discovered.
The practical measure of a good double materiality assessment is not whether it produces an elegant matrix. It is whether, eighteen months later, someone who was not in the room can open the file and see what was decided, on what evidence, by whom, and why the answer was not something else.
Sources & Verification: Every Double Materiality Figure
Every figure in this guide is listed below with the source it came from and how it was obtained. Where a number is our own arithmetic on a published figure rather than a published figure itself, the row says so — and so does the sentence in the body.
| Figure or claim | Source & status |
|---|---|
| 91 issuers · 23 Member States · about one third voluntary · all under assurance | ESMA fact-finding, 14 October 2025 — published |
| 90% of the sample above 1,000 employees · median close to 10,000 | ESMA, sample characteristics — published |
| 61.7% met the objective of IRO-1 (rounded to ~62% in the takeaways) | ESMA, key findings — published; measured on the reduced sample of 81 |
| Close to 80% explained thresholds · 17 issuers, almost 20%, disclosed none | ESMA, materiality assessment process — published |
| 70.4% disclosed input parameters | ESMA, input parameters — published; reduced sample of 81 |
| 72.5% assessed impacts gross | ESMA, gross impacts — published |
| 27.5% could not demonstrate gross assessment | ESMA, gross impacts — ⚠ derived: complement of 72.5%, labelled as such in the text |
| 93.8% consulted stakeholders · 71.6% explained adequately | ESMA, stakeholder consultation — published; 93.8% on the reduced sample of 81, rounded to “nearly 94%” in the text |
| 80.2% used due diligence to inform the assessment | ESMA, due diligence — published |
| 78% separated the impact and financial processes | ESMA, assessment process — ⚠ derived: ESMA publishes the complement, 22% did not |
| 54.3% sufficient on non-material topics · over 27% disclosed nothing | ESMA, IRO-1 topical datapoints — published |
| Close to 90% described their material IROs | ESMA, disclosure of material IROs — published |
| 71.4% targets or negative statement · 28.6% neither | ESMA, disclosure of targets — ⚠ derived: sum of 39.5% and 31.9%; ESMA’s own text says more than 71% |
| Topic materiality: E1 100 · S1 98.9 · G1 96.7 · E5 78 · S4 71.4 · S2 69.2 · E2 50.5 · E4 50.5 · S3 48.4 · E3 41.8 | ESMA, topics with at least one IRO — published; the figures behind the chart, and the source of the “100% of the sample” claim for climate |
| 26.4% disclosed an entity-specific IRO on data, cybersecurity or AI | ESMA, entity-specific matters — published |
| 2 qualified opinions · 4 emphases of matter (“four more” in the text) · one flagging a tier-1 limitation | ESMA, materiality in the assurance opinion — published |
| The 1–5 scale example: one location, a few, widespread, all | ESMA, good practice on impact thresholds — published as an ESMA-identified good practice |
| Financial thresholds anchored to revenue or EBITDA | ESMA, good practice on financial thresholds — published |
| Severity = scale, scope, irremediability with likelihood · human rights precedence · gross assessment · IRO-1 §53, IRO-2, SBM-3 | ESRS 1 and ESRS 2, Delegated Regulation (EU) 2023/2772 — legal text |
| The four steps · non-authoritative status · direct or indirect engagement · group and subsidiary guidance | EFRAG IG 1, final version May 2024 — guidance, non-binding by its own terms |
| Double materiality first formalised in 2019 | Commission Guidelines 2019/C 209/01 — legal text; the source uses the term itself |
| Mandatory datapoints cut by more than 60% · standards adopted 3 July 2026, not yet in force | European Commission, 3 July 2026 — published |
| Scope after Omnibus I, referenced for context only | Directive (EU) 2026/470 — legal text |
| Top-down route · no obligation to assess every IRO · reasonable and supportable information · the Commission’s targeted modifications | PwC, ESRS (2026) — a deep dive into the revised standards, July 2026 — ⚠ analyst: no primary link; traced to the adopted act by PwC, not read by us at source |
| Deletion of the ESRS 1 climate non-materiality paragraph | Ropes & Gray, analysis of EFRAG exposure drafts, 2025 — ⚠ analyst: traced to the exposure draft, not the adopted annex |
Read the ESMA figures for what they are. The fact-finding covers financial year 2024 double materiality statements prepared under ESRS Set 1, by a sample selected by national enforcers rather than for statistical representativeness, with early filers over-represented. Some percentages are measured on a reduced sample of 81 issuers rather than the full 91, and the ledger above says which. It describes the state of practice before the 2026 revision. We use it to show what enforcers looked for and where preparers struggled — not to predict compliance rates under the revised standards, which no one can yet measure. Two claims in this guide carry no primary link at all, because we could not read them at source: both are marked analyst in the ledger, and one of them — the fate of the ESRS 1 climate paragraph — remains an open verification on this article and on the companion guide alike.
Frequently Asked Questions
What is double materiality?
The principle that a sustainability topic must be reported if it is material from either of two perspectives: the effect the topic has on the undertaking’s financial position and performance, or the effect the undertaking has on people and the environment. Either perspective is sufficient on its own. It is the mechanism that decides which of the ten topical ESRS standards apply to a given company.
What is the difference between impact materiality and financial materiality?
Impact materiality looks outward at what the company does to people and the environment, scored on severity — scale, scope and irremediability — combined with likelihood, and assessed gross of the company’s own mitigation. Financial materiality looks inward at risks and opportunities that could affect the company’s development, position, performance, cash flows or cost of capital, scored on magnitude and likelihood. For potential negative human rights impacts, severity takes precedence over likelihood.
Is double materiality still required after the 2026 simplification?
Yes. It survived every round of the Omnibus process unchanged in principle and in structure. What changed around it is the volume of resulting disclosure and the amount of procedure prescribed — a top-down route is now permitted, exhaustive assessment of every possible impact is not required, and the previously mandated topic list became an optional appendix.
How do you set materiality thresholds under ESRS?
You set them yourself and disclose them. The standards prescribe no scoring methodology and no numeric cut-off. Common practice is a five-level scale with a defined threshold above which a topic is material, separate scales for impact and financial materiality, and — as a mark of the stronger disclosures reviewed by ESMA — financial thresholds anchored to something external such as a percentage of revenue or EBITDA, with each level of the impact scale defined in terms of the company’s own operations.
Do you have to report on topics that are not material?
Yes, in a limited but specific way. You disclose the basis for concluding that a topic is not material, and the topical standards carry their own process datapoints covering how the assessment was conducted. This was the weakest area in ESMA’s review: only 54.3% of issuers disclosed sufficient information on topics they had concluded were not material, and more than 27% disclosed nothing. Climate is the special case — its omission requires an explicit justification.
What is the difference between double materiality and single materiality?
Single materiality — the approach behind most non-EU frameworks, including IFRS S1 and S2 — asks only how sustainability matters affect the company. Double materiality adds the outward direction: how the company affects people and the environment, whether or not that ever costs it money. A company reporting under a single-materiality framework has done roughly half of what ESRS requires, and the missing half is usually the impact side of the value chain.
Is the materiality matrix required by ESRS?
No. The four-quadrant chart is a communication device, not a disclosure requirement, and it appears nowhere in the standards. What is required is the process description under IRO-1 — methodology, thresholds, input parameters, stakeholder involvement — and the description of material IROs under SBM-3. Enforcers reviewing the first cycle asked what the scale meant and who decided, not to see a matrix.
How often must the assessment be redone?
There is no fixed cycle. The revised standards ask undertakings to consider annually whether significant changes — to activities, structure, business relationships, methodology or the external environment — would alter the conclusions. In practice that means a documented annual review, with a full re-run when something material about the business has actually changed rather than on a calendar rhythm.
About This Guide
Written and maintained by the AiGreenTools Editorial Team, which covers ESG, QHSE and industrial software for buyers running procurement and compliance programmes. This is a double materiality methodology reference, not legal or assurance advice; materiality conclusions should be confirmed with your assurance provider.
| Published | 23 August 2026 |
| Last verified | 23 August 2026 — against the ESMA fact-finding report, EFRAG IG 1 and the consolidated ESRS Set 1 text, each read at source |
| Next scheduled review | 4 September 2026, when the scrutiny period on the revised standards closes |
| Evidence base | Every statistic on the first reporting cycle comes from a regulator that read the statements, not from vendor commentary about them |
| Stated limits | The ESMA sample is not statistically representative and describes practice under Set 1; we say so rather than generalising from it |
| Sourcing rule | Primary sources for legal and empirical claims; practitioner analysis labelled as such; no figure carried from another publisher’s summary without tracing it to source |
| Commercial policy | No vendor pays for placement or scoring. Methodology published at methodology |
| Corrections | contact@aigreentools.com — we publish corrections rather than editing silently |
Where to Go Next
This guide covers the double materiality method. For the framework it sits inside — what ESRS is, who must report after Omnibus I, and how ESRS differs from the CSRD — start with the companion pillar, or read What is CSRD? for the directive itself. Climate is the topic most likely to come out material for you, and the obligations that follow run through carbon accounting, climate risk and TCFD and net zero planning. If your assessment pushes materiality into the supply chain — as it did for 69.2% of the companies ESMA reviewed — supply chain ESG covers the platforms built for it. For a worked example of DMA tooling in a real platform, see our review of Greenly. Primary sources: ESMA’s fact-finding report, EFRAG IG 1 and the consolidated ESRS Set 1 text.
