In June 2026 the UK Department for Energy Security and Net Zero (DESNZ) published its annual greenhouse gas conversion factors for company reporting. The headline: the UK electricity generation factor fell by 26% compared with the 2025 set. Any company that reports Scope 2 emissions using that factor saw its electricity emissions shrink overnight without switching off a single light. A few months earlier, the European Commission’s simplified EU Taxonomy delegated act entered into force and changed which activities companies must assess. And across Europe, statistical offices and banks have been moving from NACE Rev. 2 to NACE Rev. 2.1 industry codes.
None of these are data errors. They are reference data changes: external, authoritative lookup values that feed sustainability calculations, classifications and disclosures. As ESG reporting becomes subject to assurance, regulators and auditors increasingly ask the same question finance teams have answered for decades: which version of which reference table produced this number? This article explains why emission factors, taxonomies and industry classifications should be managed as governed reference data, and how to build that capability in practice.
Why ESG Numbers Are Only as Good as Their Reference Data
Most corporate carbon footprints are calculated, not measured. Activity data such as kilowatt-hours, liters of fuel, tonne-kilometers or spend in a procurement category is multiplied by an emission factor. Classification data then decides where the result lands: which Scope 3 category, which business segment, which taxonomy activity, which sector benchmark. The activity data comes from operational systems. The factors and classifications come from outside the company, from government agencies, standard setters, industry bodies and commercial databases.
That external layer has every characteristic of classic reference data. It is owned by someone else. It changes on a published cadence. It is reused across many calculations. And small changes in it propagate everywhere. Yet in many organizations it lives in spreadsheets attached to a carbon accounting tool, copied by a consultant in a previous reporting cycle, with no record of the version, the effective date or why a particular factor was chosen.
- Emission factors: government conversion factors such as the DESNZ set in the UK and the US EPA emission factors hub, grid-average and residual-mix electricity factors, supplier-specific factors, and environmentally extended input-output factors for spend-based Scope 3.
- Sustainability taxonomies: the EU Taxonomy’s activity list and technical screening criteria, and the growing family of national and regional taxonomies modeled on it.
- Industry classifications: statistical schemes such as NACE, NAICS and ISIC, plus commercial schemes used by investors and index providers.
- Disclosure frameworks as code lists: ESRS and ISSB data points, digital taxonomy elements for tagging, and Scope 3 category definitions from the GHG Protocol.
What Changed in 2025–2026
Emission factors: methodology moves the numbers
The DESNZ 2026 major changes report explains why the UK electricity factor fell so sharply. Earlier editions used grid data two years behind the publication year; the 2026 edition shortened that lag to one year, so the update captured two years of change at once. DESNZ attributes about 16 percentage points of the drop to grid decarbonization between 2023 and 2024, about 3 points to the additional year of data, and the remaining 6 to 7 points to methodology changes, including better handling of electricity imports and exports and a fix for double counting related to autogenerators. Because electricity feeds factors for electric vehicles, rail and homeworking, those categories moved too.
The lesson for reporting teams is not that the factor is wrong. It is that a year-on-year emissions change can be driven by the reference data rather than by operational performance, and a credible report has to separate the two.
Scope 2 rules are being rewritten
The GHG Protocol is revising its Scope 2 Guidance. In July 2025 its Independent Standards Board approved a public consultation on proposals that include hourly and geographic matching for market-based clean energy claims and a hierarchy for location-based emission factors that favors more precise, regional factors where activity data and free, public factors are available. The Board signaled finalized text on key Scope 2 requirements around mid-2026, a second consultation on remaining topics, and publication of the full revised standard suite planned for the end of 2027. For data teams, that means emission factor reference data is about to become more granular: hourly, regional and contract-specific rather than one annual national average.
The EU Taxonomy got simpler, and its reference data changed
The Commission’s simplification delegated act was published in the Official Journal on 8 January 2026, entered into force on 28 January 2026 and applies from 1 January 2026. It introduces a 10% materiality threshold below which non-financial undertakings need not assess activities for alignment, and law firm summaries report that required data points fall by roughly 64% for non-financial companies and 89% for financial undertakings. Companies reporting on financial year 2025 could choose between the old and new rules, but had to state which they applied. That is exactly the kind of version choice that must be recorded alongside the numbers.
Industry codes switched versions
NACE Rev. 2.1, the updated EU statistical classification of economic activities, is being deployed under Commission Delegated Regulation (EU) 2023/137 and a series of domain regulations. Eurostat recommends harmonized implementation across banking reporting frameworks from 1 January 2026, notes that the two versions differ considerably, and points to correspondence tables and backcasting of historical series. Some statistics are reported under both versions during a transitional period. Any company or lender that uses NACE codes for sector exposure, taxonomy eligibility screening or financed emissions now needs a governed crosswalk between versions.
When an emission factor drops 26% in one release or an industry code splits in two, the question auditors will ask is not whether you used the latest table. It is whether you can prove which table you used, when, and why.
Principles for Governing ESG Reference Data
The discipline that banks apply to currency codes, holiday calendars and instrument identifiers translates directly to sustainability. Five principles do most of the work.
- Treat every external table as a versioned data set. Store the publisher, edition, publication date, effective period, license terms and checksum of the original file. Never overwrite last year’s factors; add a new version.
- Separate selection from calculation. The rule that chooses a factor for an activity, such as country grid average versus residual mix versus supplier-specific, is a governed decision with an owner and an approval history, not a formula buried in a spreadsheet cell.
- Effective-date everything. Calculations should resolve the factor and classification valid for the reporting period, so restatements and comparisons are reproducible.
- Maintain crosswalks as first-class objects. Mappings from internal product, cost center or spend categories to NACE codes, taxonomy activities and Scope 3 categories carry as much judgment as the factors themselves. They need history, rationale and review.
- Record lineage from disclosure back to source. Every reported figure should trace to the activity records, the factor version, the mapping version and the rule that combined them.
A Reference Data Architecture for Sustainability
In practice, the capability looks much like any reference data hub, with a few sustainability-specific features.
Ingestion and normalization
Factor sets arrive as spreadsheets, PDFs with annex tables, CSV downloads and commercial APIs. Each needs a loader that maps the publisher’s structure to a common model: activity type, unit, gas or CO2e basis, geography, year, source and uncertainty where provided. Units deserve special care. Kilowatt-hours versus megawatt-hours, gross versus net calorific value, and well-to-tank versus combustion factors are common sources of silent error.
Stewardship workflow
New releases should go through a review step before they become active. A steward compares the new edition with the previous one, flags factors that moved beyond a threshold, reads the publisher’s methodology notes, and documents the expected impact on reported figures. Only then is the version approved and published to calculation engines.
Distribution and consumption
Carbon accounting tools, data platforms, product footprint calculators and financed emissions models should all consume factors and classifications from the same hub, by API or governed tables in the lakehouse, rather than maintaining private copies. Where a commercial tool insists on its own factor library, the hub should at least record which library version the tool used for each reporting run.
Controls and evidence
- Automated checks that every activity record resolves to exactly one factor and one classification for its period.
- Variance reports that split year-on-year changes into activity, factor and mapping effects.
- Access controls and approval logs for changes to selection rules and crosswalks.
- A snapshot of all reference data versions used for each published report, retained with the report for assurance.
Playbook: Standing Up ESG Reference Data Management in 90 Days
- Weeks 1–2: Inventory. List every external table used in sustainability reporting: factor sets, taxonomies, classifications, framework data points. Record where each copy lives, who updates it and which reports depend on it.
- Weeks 3–4: Assign ownership. Name a business owner for each domain, typically the sustainability lead for factors, the finance or controlling team for taxonomy mapping, and risk or data governance for industry codes. Name stewards who will run releases.
- Weeks 5–7: Build the model and load history. Create the versioned data model and load at least the current and previous editions of each major source, so the first variance analysis is possible.
- Weeks 8–9: Encode selection rules and crosswalks. Move factor-selection logic and category mappings out of spreadsheets into governed tables with rationale fields.
- Weeks 10–11: Connect consumers. Point the carbon accounting tool, data platform and reporting layer at the hub. Where that is not possible, capture the tool’s factor version at every run.
- Week 12: Dry-run a release. Simulate the next annual factor update end to end, including impact analysis, approval and a variance report for management.
A Worked Example: A Manufacturer Explains Its Scope 2 Drop
Imagine a hypothetical manufacturer with plants in the UK, Germany and Vietnam that reports under ESRS. Its FY2026 location-based Scope 2 emissions fall noticeably compared with FY2025, and the board is pleased. The sustainability team is cautious, because energy consumption barely changed.
With a governed factor hub, the variance analysis takes an afternoon. The team reruns FY2026 activity data against the FY2025 factor versions and isolates the effect. Most of the UK reduction comes from the new DESNZ electricity factor rather than from efficiency projects, and the methodology report explains why. The German and Vietnamese figures move for their own reasons: updated grid factors from the relevant national sources and a newly signed supplier contract that changes the market-based result. The disclosure narrative separates operational improvements from reference data effects, and the assurance provider can trace each number to a dated, approved factor version.
The same hub lets the finance team rerun taxonomy eligibility using the 2026 simplified rules and the 10% materiality threshold, while keeping the FY2025 comparison on the option the company had chosen. Meanwhile, the risk team uses the NACE Rev. 2 to Rev. 2.1 crosswalk to keep sector exposure reporting consistent for a lender that requested the new codes. Without versioned reference data, each of these questions would have triggered a manual reconstruction exercise.
Common Pitfalls
- Silent factor updates inside tools. Some platforms refresh their libraries automatically. Without a record of the version used, last year’s report becomes unreproducible.
- Mixing factor vintages. Combining a new electricity factor with an old transport factor set in the same period produces internally inconsistent totals.
- One-way mappings. Mapping spend categories to Scope 3 categories without the reverse view makes it hard to explain where a figure came from.
- Ignoring license terms. Commercial factor databases and some classification schemes restrict redistribution. The hub must enforce who can see and export what.
- Treating crosswalks as IT artifacts. A mapping from an internal product family to a taxonomy activity is a reporting judgment. It needs the same review as an accounting policy.
- Waiting for the final standard. Scope 2 revisions and taxonomy refinements will keep coming. A versioned model absorbs them; a spreadsheet does not.
Outlook: Granular, Frequent and Assured
The direction of travel is clear. Emission factors will become more granular as Scope 2 moves toward regional and hourly data. Digital tagging of sustainability statements means classification and data point code lists will be machine-read by regulators and investors. Assurance requirements under CSRD and similar regimes will push companies toward the same reproducibility that financial reporting already demands. And AI assistants that draft sustainability narratives or answer questions about emissions will only be trustworthy if they can cite the factor and mapping versions behind each figure.
Reference data management is not glamorous, but it is the mechanism that turns a pile of external tables into an auditable sustainability data supply chain. Companies that build it now will spend less time reconstructing last year’s numbers and more time explaining real progress.
Conclusion
The 2026 cycle delivered a clean demonstration of the problem: a large factor change driven partly by methodology, a simplified taxonomy with a transitional choice, and an industry classification switch with dual reporting. Each one is manageable on its own. Together, without governance, they make sustainability figures hard to compare and harder to assure. Start by inventorying your external tables, give each one an owner and a version history, move selection rules and crosswalks into governed data, and make every published number traceable to the reference data that produced it.
How Hanoi Technology Can Help
Hanoi Technology builds versioned reference data hubs for emission factors, taxonomies and industry classifications, connected to your data platform and reporting tools so every ESG figure is traceable.
Further Reading
- 2026 GHG conversion factors: major changes report – UK DESNZ
- DESNZ release 2026 UK GHG conversion factors – Circular Ecology
- Scope 2 standard advances: ISB approves consultation – GHG Protocol
- Omnibus update: EU Taxonomy delegated act enters into force – Herbert Smith Freehills Kramer
- NACE transition – Eurostat