Many companies and investors want a simple benchmark to judge whether a greenhouse gas (GHG) inventory is “reliable” or “complete.” In practice, quality cannot be guaranteed by citing a single framework or standard, it depends on how thoroughly the methods are applied, how transparent the boundary-setting is, and how credible the underlying data and governance are.
Automated or AI-assisted carbon calculators have made GHG accounting more accessible, but not necessarily more accurate. Many of these tools include disclaimers such as “AI can make errors; use with caution”, reminding users that automation does not replace professional judgment.
Likewise, when companies calculate in-house or through consultants, they commonly claim the inventory was prepared “with reference to” or “in accordance with” the GHG Protocol. From a professional standpoint, such phrases do not automatically guarantee robustness nor mean it meets the Protocol’s full intent.
GreenCo, an ESG advisory firm established in 2016, shares the following professional guidance for assessing the quality of Scope 3 disclosures, particularly in an era when auto-generated numbers risk becoming the norm.
1) Start with relevance screening: which of the 15 categories are covered?
Scope 3 emissions are divided into 15 categories, 8 upstream and 7 downstream. The GHG Protocol requires companies to screen all categories for relevance based on criteria such as:
- Size – the expected magnitude of emissions,
- Influence – the company’s ability to affect or reduce those emissions,
- Risk – exposure to regulatory, market, or reputational risks,
- Stakeholder interest, and
- Data availability and quality.
A credible disclosure should clearly explain:
- Which categories are included and which are excluded,
- The rationale for each decision, and
- The screening results, supported by screening logic, thresholds, and evidence
- Avoid generic blanket statements like “excluded because it is immaterial”
Red flags
- Categories marked “not relevant” despite clear relevance (e.g., a retailer excluding Purchased goods and services; a food and beverage player [SC1] excluding Waste generated during operations).
- Categories excluded with no documented screening analysis or rationale for exclusions.
- Double counting or obvious omissions (e.g., reporting Use of sold products for a fuel company as “N/A”).
Acceptable exclusions—if justified
It can be reasonable to exclude categories that are demonstrably not relevant for the business model. The key is evidence: a screening showing that the category’s estimated emissions are trivial in magnitude and do not affect decision-making or target-setting.
2) Review boundary and scope definitions
After identifying relevant categories, the next quality indicator is whether the organizational and operational boundaries are properly defined and applied.
Professionals should look for:
- Organizational boundary: Whether the inventory follows an equity share, financial control, or operational control approach, and applies consistently to subsidiaries, JVs, and franchises.
- Value chain boundary: Whether relevant upstream and downstream partners are included in the scope of data collection.
- Reporting period and currency consistency: Whether the reporting year matches activity data years, and whether financial data (used for spend-based methods) are adjusted for exchange rates and inflation appropriately.
- Avoided emissions: Whether these are kept separate, clearly labelled, and not netted against the scope 3 totals.
3) Assess methodological robustness, beyond the name of the standard
Citing the “GHG Protocol” is the beginning, not the end. A professional review should examine how rigorously the methodology has been implemented. Key elements include:
- Data hierarchy - Evaluate the share of primary vs. secondary data:
- Best: Supplier-specific primary data (measured, metered or process-level data).
- Good: Activity-based secondary data (e.g., physical quantity × emission factor).
- Basic: Spend-based estimates (monetary) using sector averages.
- Emission factors: Checked named sources, versions numbers, geographies, and publication years (e.g., DEFRA/BEIS 2024, ecoinvent xx.x).
- Allocation rules: For multi-output processes, whether allocation follows consistent and transparent rules (mass, energy, or economic value basis)
- Sampling: If sampling is used (e.g. for suppliers or sites), whether the documented methodology defines sample frames, representativeness, confidence intervals, and scaling logic.
- Updates and restatements: If methods or emission factors are updated, whether prior-year data are restated with explanation.
4) Scrutinize exclusions and justifications
The quality of a disclosure depends as much on what is excluded as on what is included. When reviewing justifications, consider:
- Relevance screening evidence: Has the company documented why certain categories were deemed not relevant, using the GHG Protocol criteria?
- Counter-examples: If you can identify obvious activity contradicting the exclusion, raise a red flag.
- Resource limitation claims: If the company cites limited resources or data constraints, assess whether this is a temporary technical challenge or a management choice.
- For financially strong companies, repeated use of “resource limitation” may indicate lack of prioritization rather than actual constraint.
- Confidentiality claims: Rarely valid, except for national security or highly sensitive trade contexts. In most cases, aggregated or anonymized disclosure should still be possible.
5) Industry-specific quality signals (examples)
Each sector has its own Scope 3 risk profile. Reviewers should watch for sectoral alignment, for example:
- Apparel and retail: Category 1 (Purchased goods and services) usually dominates; credible inventories should include supplier-specific data and chemistry-stage impacts.
- Food and beverage: Agricultural emissions (land use change, enteric fermentation) and upstream logistics matters are key; biogenic CO₂ and CH₄ treatment should be clearly disclosed.
- Technology hardware: Capital goods and upstream manufacturing are often major sources; check regional electricity factors and yield scrap treatment.
- Oil, gas and energy: Downstream “Use of sold products” is typically material; exclusions here require very strong justification.
- Financial institutions: Financed emissions are outside corporate Scope 3 but often expected; where disclosed, look for PCAF-style methods and data quality flags.
6) Human oversight in the age of automation
AI and automation can assist data collection and estimation but cannot replace professional judgment. Companies using automated tools should demonstrate:
- Human verification of AI outputs
- Error-checking procedures
- Documentation of assumptions and decision logic
Automation without oversight risks producing numbers that appear “precise” but are methodologically weak.
Conclusion
High-quality Scope 3 reporting is not about perfection, nor citing a standard alone. It is about transparency, credibility, and continuous improvement. A Company earns trust when it can clearly explain what is covered, how relevance was determined, what data sources were used, and how it plans to improve over time.
If a company starts with partial coverage (e.g., selected sites or supplier samples), that’s acceptable only with a clear plan, milestones, and visible year-over-year improvement.
Scope 3 disclosures, when done well, is not just a compliance task, but a strategic management tool that links supply chain, investment, and innovation decisions with climate strategy.
Appendix: standard questions for companies
- Which categories are excluded and why? Please share your quantitative screening.
- For Category 1, what % of spend/volume is covered by supplier-specific data?
- What are your primary emission factor sources and versions?
- How do you allocate multi-output processes?
- Did methods/factors change since last year? What was restated?
- What controls and QA steps do you use? Is any part externally assured?
- How are automation/AI outputs reviewed by humans?
- What is your 2-year plan to increase primary data coverage and supplier participation?