Data collection is typically the most time-consuming part of building a Scope 3 inventory. The standard provides a data quality hierarchy that guides companies on what types of data to seek, in order of preference. The hierarchy reflects a fundamental principle: more specific data produces more accurate emissions estimates.
The Data Quality Hierarchy
The standard ranks five types of data from highest to lowest quality:
1. Primary Data (Highest Quality)
Data obtained directly from specific activities in the value chain — either the company's own activities or those of specific supply chain partners.
Examples:
- Energy consumption records from a specific supplier's factory (for Category 1)
- Fuel receipts from a third-party logistics provider for shipments to the company (for Category 4)
- Employee travel data from a corporate travel booking system (for Category 6)
- Energy bills for leased facilities (for Category 8 or 13)
Primary data is the most accurate because it reflects the actual conditions at the specific facility, at the specific time, with the specific technology and energy mix in use.
2. Secondary Data: Industry-Average Data
Data based on averages across a sector or industry group. Published by:
- Life cycle assessment (LCA) databases (ecoinvent, GaBi, SimaPro)
- Government statistical agencies
- Industry associations (World Steel Association, Cement Sustainability Initiative)
- Academic peer-reviewed lifecycle studies
Industry-average data is less accurate than primary data but more accurate than economic input-output data, because it reflects physical production processes rather than economic relationships.
3. Secondary Data: Proxy Data
Data from similar but not identical activities, adjusted to approximate the specific case. For example, using the emission factor for a nearby geography when the specific geography's factor is unavailable, or using a similar product's LCA when the exact product LCA is absent.
4. Secondary Data: Economic Input-Output (EIO) Data (Lowest Quality)
Data derived from national economic input-output models, expressing emissions per unit of economic output for broad industry sectors. The foundation of the spend-based method — multiply spend by an EIO emission factor.
EIO data is the least accurate because:
- It aggregates across very different companies within a sector
- It uses monetary flows rather than physical quantities
- It does not distinguish between emission intensities of individual suppliers
The hierarchy is not a rigid requirement — companies should use the best data available for each category and activity, moving toward higher-quality data over time for the most significant categories. Using EIO data for a minor purchased service category is entirely acceptable. Using EIO data for a major raw material category that represents 40% of Scope 3 emissions is not a long-term solution.
Collecting Data in Practice
Supplier Data Collection
For Category 1, collecting primary data requires engaging suppliers. The standard recommends:
- Identify top suppliers: Rank by annual spend; the top 10–20 suppliers often represent 50–80% of total purchasing spend.
- Assess data availability: Check if key suppliers already participate in CDP or publish sustainability reports with Scope 1/2 data.
- Request data: Use structured questionnaires or integrated supply chain platforms (e.g., EcoVadis, Sedex, Supplier.io) to request emissions data.
- Validate responses: Check for internal consistency; compare reported emission intensities against industry benchmarks.
- Use for calculation: Convert supplier-provided Scope 1/2 data to per-unit emission factors, then multiply by quantities purchased.
Activity Data Sources by Category
| Category | Typical Activity Data Source |
|---|---|
| 1 — Purchased Goods | Procurement records, supplier invoices, purchase orders |
| 4 — Upstream Transport | Logistics bills of lading, shipping manifests, freight invoices |
| 5 — Waste | Waste contractor manifests, waste audits, facility waste records |
| 6 — Business Travel | Travel management company (TMC) reports, expense claims |
| 7 — Employee Commuting | Employee travel surveys, HR headcount data |
| 11 — Use of Sold Products | Product specifications, energy ratings, sales volume data |
Data Quality Assessment
The standard recommends assessing data quality across five dimensions (adapted from the ecoinvent data quality framework):
- Technological representativeness: Does the data reflect the technology actually used?
- Temporal representativeness: Is the data from the same time period as the reporting year?
- Geographical representativeness: Does the data reflect the specific geography of the activities?
- Completeness: Does the data cover all relevant flows and processes?
- Reliability: Is the data from a credible, verifiable source?
Data quality assessment is like checking a recipe against what you actually cooked. A recipe (industry-average data) gives you a reasonable idea of the dish's nutritional content. But if you actually weigh every ingredient as you cook (primary data), you know exactly what's in your specific meal. The standard says: use the recipe estimate when cooking occasionally, but invest in scales when cooking is your main business.
Estimating and Addressing Data Gaps
Not all activity data will be available. The standard allows companies to use estimation techniques for data gaps:
- Extrapolation: Use known data from a subset of operations to estimate the whole (e.g., survey 20% of employees and extrapolate to the full workforce)
- Interpolation: Fill gaps between known data points using logical assumptions
- Engineering estimates: Use process knowledge to estimate emissions from physical principles
- Proxy data: Substitute a similar activity's data when the specific data is unavailable
All estimations and gaps should be documented, with the basis for the estimate recorded, to support assurance and year-on-year consistency.
Traditional Scope 3 data collection is manual and labour-intensive. Increasingly, companies are using digital supply chain platforms that integrate with ERP systems to automatically extract procurement volumes, freight data, and waste records. AI-powered tools can match purchase categories to emission factors, flag anomalies, and automate supplier outreach workflows. The EU's CSRD-mandated sustainability data requirements are accelerating investment in these platforms, as companies need auditable Scope 3 data at scale. However, as of 2024, primary supplier data collection remains challenging, and spend-based estimates still dominate Category 1 for most companies outside the Fortune 500.
Key Takeaways
- 1The data quality hierarchy ranks sources from highest to lowest: primary supplier data, industry-average data, proxy data, and economic input-output (EIO) data
- 2Focus primary data collection on top suppliers - the top 10-20 by spend often represent 50-80% of total procurement
- 3Assess data quality across five dimensions: technological, temporal, and geographical representativeness, completeness, and reliability
- 4Data gaps can be addressed through extrapolation, interpolation, engineering estimates, or proxy data - but all estimation methods must be documented
- 5Use EIO (spend-based) data for minor categories but invest in higher-quality methods for categories representing the largest share of Scope 3 emissions