ESRS Digital Tagging and XBRL: What Preparers Need to Build Now

Technical status. The Commission adopted revised ESRS on 3 July 2026, but its implementing page continued to state that the amending delegated regulation was not in force until publication in the Official Journal. EFRAG’s existing Set 1 XBRL taxonomy was developed for the 2023 ESRS. ESMA reported in May 2026 that it had temporarily paused finalising digital sustainability-reporting requirements while legislative and technical developments were being resolved. Recheck all three points before publication.

Educational material. It does not replace the applicable delegated act, national law, legal advice, professional judgement or an assurance conclusion.

Digital reporting is often treated as a final-stage conversion exercise: finish the sustainability statement, send a PDF or Word file to an XBRL vendor, and ask the vendor to tag it. That approach transfers unresolved reporting decisions into the tagging phase. It also produces avoidable questions about which sentence contains the fact, which boundary applies, whether a number is an instant or duration fact, whether a narrative block should be tagged, and whether a company-specific disclosure requires an extension.

A better model is to treat digital reporting as controlled information architecture. The same record that supports drafting should state what the requirement is, why it applies, where the disclosure appears, which evidence supports it, how the fact is defined and which taxonomy concept is the current mapping candidate. That does not mean tagging a report prematurely. It means avoiding a reporting system in which human-readable text, data workbooks and future machine-readable facts evolve independently.

The commercial and assurance consequences are practical. Poor mapping can expose inconsistencies that were hidden in prose. A validation engine can detect certain structural defects, but it cannot decide whether a material impact was omitted, whether an estimate is reasonable or whether a policy description is balanced. Preparers therefore need both technical controls and reporting judgement.

The first layer is the legal obligation to prepare and publish the sustainability statement. It determines the reporting undertaking, period, location in the management report, applicable standards, approval and assurance. Digital tagging does not determine whether a disclosure is legally required; the legal and materiality analysis comes first.

Article 29d of the Accounting Directive provides the legal basis for electronic reporting and marking up sustainability information. Directive (EU) 2026/470 added an important practical safeguard: undertakings are not required to mark up sustainability reporting until the relevant delegated rules are adopted. This distinction should appear prominently in project plans and board papers.

The human-readable statement is what management, the board, users and the assurance provider read. It must present the required information coherently, with suitable structure, cross-references and connected information. A technically valid tag cannot rescue an ambiguous boundary, a missing methodology or a generic policy statement.

A taxonomy is a controlled dictionary and relationship model. It contains concepts, labels, data types, references, dimensions and other technical relationships. EFRAG’s ESRS Set 1 taxonomy provides elements for datapoints and dimensional disaggregation in the 2023 ESRS. It is an important source for understanding the future fact model, but it is not itself the Commission’s final filing rule.

The tagged filing combines visible content with machine-readable facts, typically through Inline XBRL within XHTML. It also depends on an entry point, technical specifications, filing rules, permitted extensions, validations and submission arrangements. These elements are final only when the applicable regulatory package is adopted and becomes effective.

Figure 1. Current reporting, preparer readiness and the future mandatory digital layer. Original London Reporting Academy practitioner visual.

The most valuable preparer work is taxonomy-neutral enough to survive a change in taxonomy version, but structured enough to support rapid remapping.

For each ESRS disclosure requirement and material datapoint, record:

a stable internal requirement ID;

the standard, disclosure requirement and source edition;

the materiality and applicability conclusion;

the relevant impact, risk or opportunity;

the reporting boundary and period;

the report location;

the data or narrative owner;

evidence and methodology;

review status and final approval;

the current taxonomy mapping candidate;

any extension or dimensional issue;

the taxonomy and mapping version.

The register should not be a static content index. It is a control object connecting the requirement, published text and supporting evidence.

Do not make the taxonomy concept name the only identifier in the source system. A taxonomy can change, a concept can be renamed or a revised standard can combine or remove requirements. The internal datapoint ID should remain stable while the mapping history records the external concept used for each reporting cycle.

A useful ID does not need to reproduce the legal numbering. For example, CLIMATE.GHG.S2.LOCATION.GROSS can remain stable even if the external taxonomy concept or disclosure numbering changes. The record then stores the relationship to E1-8 and the relevant taxonomy concept for the selected edition.

A numerical value is not a complete machine-readable fact. Preparers should define, as applicable:

data type: monetary, decimal, percentage, integer, Boolean, date, enumeration, text or text block;

period: instant or duration;

reporting entity and consolidation scope;

unit and scaling;

geography, site, business segment or other dimensions;

gross or net basis;

current period, comparative period and restatement status;

estimate status and methodology version;

whether the visible disclosure is narrative, semi-narrative or quantitative.

The model should also explain when a disaggregation is presented in the report but not stored as a separate source-system field. That gap often becomes a tagging and assurance issue later.

ESRS contains substantial narrative content. A digital taxonomy may include text-block concepts, Boolean confirmations and enumerated values as well as numeric facts. Preparers should therefore structure narrative drafting around controlled components: policy scope, governance responsibility, action, resource, target, methodology, limitation and progress.

This does not mean writing robotic text for a taxonomy. It means making the narrative traceable. One approved paragraph may support several machine-readable elements, and one element may require careful identification of the exact visible text that represents it.

The mapping workflow should begin only after the disclosure conclusion and wording are sufficiently stable.

A spreadsheet column called “water consumption” is not enough. It could refer to a consolidated total, a figure for water-stress areas, a site-level value, a current-period amount, a comparative, or an intensity. The final mapping depends on the approved disclosure and the contexts attached to the fact.

An extension is an entity-specific taxonomy concept used when a standard concept does not faithfully represent the reported information. Extensions can be necessary, but they can also reduce comparability and create technical errors.

Preparers should establish an extension policy with the following hierarchy:

Search the relevant taxonomy and labels thoroughly.

Test whether an existing concept represents the same accounting or sustainability meaning, not merely similar words.

Check whether dimensions or disaggregation solve the issue without a new concept.

Document why broader, narrower or alternative concepts are not faithful.

Create an extension only under the final permitted rules.

Define anchoring or relationships required by those rules.

Approve the extension through reporting and technical review.

Reassess it when the taxonomy changes.

A company-specific metric does not automatically require an extension. If it appears only as supporting information and is not required to be tagged under the final filing rules, no extension may be needed. Conversely, a required material disclosure cannot be left untagged merely because the preparer has not designed extension governance.

A mature digital record identifies at least five versions:

legal source version - the directive, delegated regulation and national implementation used;

ESRS version - the exact standards and corrigenda applied;

taxonomy version - package, entry point and release date;

mapping version - the approved relationship between internal datapoints and taxonomy concepts;

report version - the final visible management report or sustainability statement.

The linkage matters. A mapping prepared against the 2023 ESRS taxonomy cannot silently be carried into a report prepared under revised ESRS. The record should show whether the mapping was retained, changed, split, merged, removed or replaced by an extension candidate.

Use a change log that states the trigger, affected facts, owner, review, comparative implications and whether a previously approved filing must be regenerated. Store the actual taxonomy package, not only a web link that may later point to a newer version.

Technical validation can be powerful. Depending on the final rules and software, it may identify:

invalid taxonomy or schema references;

missing or inconsistent contexts;

incorrect units or scaling;

duplicate facts with inconsistent values;

invalid enumerations;

calculation inconsistencies;

missing facts flagged by formula rules;

incorrect use of extension concepts or relationships;

facts that do not correspond to visible text;

filing-package and XHTML defects.

It cannot, by itself, determine:

whether the undertaking is in legal scope;

whether the double materiality assessment is robust;

whether a topic or datapoint contains material information;

whether a boundary or estimate is reasonable;

whether a policy is actually in force;

whether actions have the stated resources;

whether a target is credible or performance is balanced;

whether current and anticipated financial effects are complete;

whether the narrative is fair, neutral and understandable;

whether a technically valid extension preserves comparability.

A clean validation report is therefore evidence of technical processing, not a compliance certificate.

Figure 2. Mapping workflow and the boundary between technical validation and human reporting review. Original London Reporting Academy practitioner visual.

The digital layer should be included in the reporting control environment, even before mandatory tagging begins.

The person who selects the taxonomy concept should not be the only person approving the mapping. The reviewer should understand both the disclosure meaning and the technical model. For high-risk facts - such as GHG emissions, financial effects, workforce incidents or EU Taxonomy KPIs - consider joint review by the subject owner, reporting team and XBRL specialist.

A final control should compare:

the value displayed in the report;

the value in the calculation file;

the value in the XBRL fact;

the context, unit, scale and sign;

the comparative and restatement treatment;

any rounding difference.

Narrative and text-block tags require a similar test: the tagged text should be visible, complete and not extend beyond the intended disclosure.

Reviewers should inspect the filing in a human-readable renderer. Technical extraction alone can miss display problems, hidden text, duplicated paragraphs or tags attached to the wrong visible content. Retain the rendered output, validation results, mapping register, extension approvals and final checksum as part of the publication evidence pack.

Context. Meridian Components prepares an ESRS sustainability statement and expects that digital tagging will become mandatory in a future reporting cycle. The revised ESRS and final ESEF taxonomy are not yet part of its effective filing requirements.

Decision. Meridian does not publish a voluntary filing described as “the mandatory ESRS XBRL report”. Instead, it creates an internal digital mapping register.

Implementation. The register connects each approved disclosure to internal datapoint IDs, report pages, owners, evidence, fact types and the current EFRAG Set 1 concept where a reasonable candidate exists. Mappings affected by revised ESRS are marked “provisional - remap after final taxonomy”. Company-specific metrics are recorded as possible extension candidates but are not implemented until the final extension rules are known.

Control. The reporting team tests the most complex disclosures in a sandbox and records technical warnings. Management receives a readiness report that separates content gaps from technical defects.

Limitation. Meridian’s exercise does not establish that the mapping will be accepted under future filing rules. It reduces migration work and reveals weaknesses in the underlying disclosure system.

Illustrative wording - adapt to facts and legal requirements. “The sustainability statement has been prepared in a human-readable format in accordance with the reporting requirements applicable to the Group for the year ended 31 December 20X6. Mandatory sustainability mark-up did not apply to this reporting cycle. The Group maintained an internal datapoint and mapping register to support future digital reporting readiness. The register used the EFRAG ESRS Set 1 taxonomy as a technical reference where relevant; mappings affected by revised standards or future ESEF rules remain provisional and are not presented as a regulatory filing.”

This wording works only when it is factually accurate. It distinguishes the report’s legal status from preparer readiness, names the provisional nature of mapping and avoids an unsupported compliance claim.

Calling EFRAG’s taxonomy a current mandatory filing taxonomy. Correct by stating its technical role and checking whether the Commission has adopted the relevant ESEF rules.

Mapping before materiality and applicability are approved. Correct by linking mapping to the disclosure decision register.

Treating PDF page extraction as a fact model. Correct by defining period, entity, unit, boundary, dimensions and estimate status.

Using the taxonomy as a data dictionary without internal IDs. Correct by separating stable internal datapoints from external taxonomy versions.

Assuming all narrative content will be tagged as one block. Correct by understanding text blocks, semi-narrative elements and final marking-up rules.

Creating extensions for convenience. Correct by documenting concept searches, dimensional alternatives and the final regulatory extension criteria.

Ignoring comparative and restatement logic. Correct by versioning contexts and reconciling changed methods or boundaries.

Letting the XBRL vendor make substantive reporting decisions. Correct by retaining management responsibility and technical-review gates.

Closing all issues because the validation report is green. Correct by maintaining separate technical and substantive findings.

Failing to regenerate after the final report changes. Correct by locking the final source file, mapping version and checksum together.

Myth: “Once the XBRL file validates, the sustainability statement is compliant.”

Reality: Validation tests the rules encoded in the taxonomy, filing specification and software. ESRS compliance also depends on legal scope, materiality, faithful representation, boundaries, methodologies, evidence, presentation and governance. A valid instance can still tag an incomplete or misleading report.

Primary legal and technical connections

Accounting Directive, Article 29d - electronic format and marking-up basis, as amended.

Directive (EU) 2026/470 - current amendments, including the no-mark-up-until-rules safeguard.

EFRAG ESRS Set 1 XBRL taxonomy and explanatory materials - technical taxonomy for the 2023 ESRS.

ESMA ESEF work - proposed marking-up rules, extensions, validation and filing mechanics.

Revised ESRS - the target human-readable requirement set that will require a corresponding digital update.

Next practical materials

ESRS Disclosure Matrix and Content Index: mapping requirements to report locations and evidence.

ESRS Data Governance and Internal Controls: datapoint owners, calculations and review controls.

ESRS Compliance Checklist and Free Disclosure Matrix: publication control tool included in this package.