Devera LCA/PCF Methodology

The carbon footprint assessment conducted by Devera follows a “Cradle to Grave” or “Cradle-to-Gate” Life Cycle Assessment (LCA) approach, focused on the climate change impact category (Global Warming Potential). The assessment can optionally be expanded to the 16 EF v3.1 (PEF) impact categories via Full LCA (see section 9). This is a screening-level LCA conforming to ISO 14040:2006 and ISO 14044:2006. The assessment evaluates greenhouse gas emissions throughout a product's entire life cycle, from raw material extraction to final disposal, and comprises the following key stages:

1. Goal and Scope

The LCA calculation aims to approximate the environmental impact of products with these objectives:

1. Identify environmental impacts.
2. Detect critical areas to optimize the use of resources, energy, and emissions, promoting more efficient production.
3. Drive the design of more sustainable products from conception.
4. Facilitate the selection of more sustainable materials, processes, and suppliers.
5. Ensure compliance with national and international environmental regulations.
6. Generate reliable reports to support environmental claims and ecological certifications.
7. Communicate information to different stakeholders.

Functional Unit

The functional unit is defined as one unit of finished product as sold to the consumer (e.g., one 250 ml bottle of shampoo, one 500 g package of pasta). All inputs, outputs, and environmental impacts are referenced to this unit, allowing consistent comparison between products within the same category.

Alongside the result per functional unit, the report also provides a mass-normalized figure (kg CO₂e per kg of product) to support comparison between products of a similar type.

System Boundaries

The “Cradle to Grave” approach covers all product phases: production, use, and end of life. The included stages are:

1. Raw materials: Encompasses the extraction, processing, and transport of raw materials used in manufacturing.
2. Packaging: Analyzes the environmental impact of packaging materials, their production, and subsequent waste management.
3. Manufacturing: Includes all industrial processes of transforming raw materials into the final product, considering waste, energy consumption, and direct emissions.
4. Transport: Evaluates emissions generated in product distribution from factory to point of sale, including land, sea, and air transport.
5. Use: Examines emissions during the product's useful life, including energy consumption and resources in operation and maintenance.
6. End of life: Evaluates emissions in the final phase, including recycling, incineration, or landfill disposal.

The following elements are excluded from the system boundaries:

• Construction and maintenance of manufacturing infrastructure and capital goods.
• Employee commuting and administrative operations.
• Retail operations at point of sale.

The “Cradle-to-Gate” approach includes the first 3 phases and excludes Transport, Use, and End of Life.

Cut-off Criteria

Per ISO 14044 §4.2.3.3: all materials representing 1% or more of total product mass are always included. Materials below 1% are also included whenever data is available. All energy inputs to manufacturing are included in full via Ecoinvent process data or published industry benchmarks, regardless of their relative share.

2. Life Cycle Inventory

Quantitative data on inputs (raw materials, energy, water) and outputs (emissions, waste) are collected at each life cycle stage.

Each material input, process, or energy source is converted to CO₂-equivalent (CO₂eq) emissions using emission factors from recognized databases. To ensure broad coverage and accuracy, Devera uses a multi-source cascade system that selects the most appropriate database depending on the product category:

• INCI Database. Primary source for cosmetic ingredients, with 667 verified emission factors from peer-reviewed studies specific to the personal care industry.
• AGRIBALYSE 3.2. Primary source for food products. Contains ~2,500 French food product LCA datasets developed by ADEME (French Agency for Ecological Transition).
• Ecoinvent 3.9.1. Primary source for all other materials. ~21,200 process datasets covering materials, energy, transport, and waste treatment worldwide.
• Base IMPACTS 3.0. ADEME (France) LCA database for consumer products. Covers textiles (cashmere yarn, angora, recycled fibres, etc.), leather (tanning, finishing), construction boards (MDF, HDF, plywood) and finished cosmetic products (shampoo, soap, detergent). Fills gaps where Ecoinvent lacks specific data.
• DEFRA 2025. UK Government emission factors (~250 factors) used as a secondary fallback for plastics, metals, textiles, paper, glass, transport, and energy.
• Proxy estimation. Final fallback when no specific factor is found in any database. A proxy factor is applied based on similar materials or industry averages. These cases are clearly flagged as “Estimated” in the report.

The cascade priority depends on the product category: for cosmetics, the INCI database is consulted first; for food products, AGRIBALYSE takes priority; for all other categories, Ecoinvent is the primary source, followed by Base IMPACTS for specialized materials (textiles, leather, boards). When a specific emission factor is not available in the primary database, the system automatically falls back to the next available source, ensuring every material receives a data-backed estimation.

All databases are periodically updated to incorporate the latest available versions. The data vintage for each source is noted alongside the emission factor in the final report, allowing users to assess temporal representativeness.

Biogenic Carbon

In accordance with ISO 14040/14044, biogenic carbon (CO₂ absorbed and released by biological materials such as wood, cotton, or food ingredients) is reported separately from fossil carbon emissions when the underlying data source provides this distinction. For food and bio-based products, the emission factors used (notably from AGRIBALYSE and Ecoinvent) account for land use and land use change where applicable.

All emission factors are evaluated using the IPCC AR6 Global Warming Potential (GWP100) methodology, which assigns a CO₂-equivalent value to all relevant greenhouse gases over a 100-year horizon.

3. Impact Assessment

The carbon footprint measures greenhouse gas (GHG) emissions associated with a product throughout its complete life cycle. For each stage, the total impact is calculated as:

CO₂eq = Material quantity × Emission Factor

The sum across all stages and materials yields the product's total carbon footprint.

Each emission factor is accompanied by a confidence level (High, Medium, Low, or Estimated) based on the specificity and quality of the data source, providing transparency about the reliability of each data point.

In addition to the per-factor confidence level, each report quantifies the share of primary data (values taken directly from brand or supplier documents) versus secondary data from published databases. This primary-data share is shown in the report and in all exports (PDF, Excel, CSV) as a data quality transparency indicator, consistent with the ISO 14044 data quality requirements.

4. Interpretation

Results are evaluated to identify potential improvements, validate objectives, and communicate findings. In the “Cradle to Grave” approach, the analysis highlights the stages and materials with the highest environmental impact, enabling targeted reduction strategies.

5. Additional Principles

1. This analysis follows ISO 14040:2006 and ISO 14044:2006, ensuring a structured, transparent, and reproducible approach at all stages.
2. Emission factors are selected with geographic specificity when available (regional data preferred over global averages).
3. All calculations and data sources are documented in an audit trail for full traceability.

6. Limitations and Uncertainty

As with any LCA based on secondary data, this methodology is subject to inherent limitations that users should be aware of:

• Secondary data: The assessment relies on emission factors from published databases rather than primary measurements from the specific manufacturer. Actual emissions may vary depending on production conditions.
• Data gaps: When a specific material or process is not found in any database, proxy estimations based on similar materials or category averages are used. These cases are clearly flagged in the report.
• Geographic approximation: When country-specific emission factors are not available, global or regional averages (e.g., GLO, RER) are used as substitutes.
• Use phase assumptions: For products where consumer behavior significantly affects environmental impact (e.g., appliances, clothing care), standardized usage scenarios are applied based on industry references.
• Uncertainty range: Individual emission factors carry an inherent uncertainty that is not propagated quantitatively through the calculation. The confidence level assigned to each factor provides a qualitative indication of data reliability.

These limitations are consistent with screening-level LCA approaches as defined in ISO 14040/14044.

7. Technology for Automating the LCA Measurement Process

Devera implements advanced technology to automate LCA, highlighting:

• Extraction and analysis of data from product pages, ingredient lists, and unstructured documents from brands and suppliers.
• AI-driven identification of the most appropriate emission factor for each material, using a multi-source database cascade tailored to the product category.
• Automated calculation of environmental impact across all six lifecycle phases.
• Obtaining representative market benchmarks to model distribution by categories and make precise comparisons.
• Complementing missing data with validated market standards and proxy estimations.
• Post-analysis verification: users can refine individual emission factors through an AI-assisted review process that searches across all available databases.

8. Impact Score Calculation

The Impact Score rates a product's carbon footprint (A–E) against a simulated market distribution for its product category, following ISO 14040/14044 principles.

How it works

1. Product classification: The product is classified into a functional subcategory (e.g., “t-shirt”, “shampoo”, “smartphone”) that groups products serving the same purpose, regardless of brand or material variant.
2. Market distribution: For each subcategory, a market distribution is built using Monte Carlo simulation (N=10,000). The simulation models six lifecycle phases — raw materials, packaging, manufacturing, transport, use phase, and end of life — using probability distributions calibrated with:
   • Published Environmental Product Declarations (EPDs) found via web search
   • Emission factors from Ecoinvent 3.9.1, AGRIBALYSE 3.2 and Base IMPACTS 3.0 (ADEME), DEFRA 2025, and peer-reviewed literature
   • Market share data for materials, manufacturing locations, and disposal routes
3. Scoring: The product's CO₂e (normalized per functional unit — per unit, per kg, or per L) is placed within the simulated distribution. The score corresponds to its percentile position:

4. Reusability: Market distributions are generated once per subcategory and reused across all products in that category, ensuring consistency and reducing computation time.

Scope handling

• For cradle-to-grave products, scoring uses the full 6-phase distribution.
• For cradle-to-gate products, scoring uses only the first 3 phases (raw materials, packaging, manufacturing), derived from the same simulation.

Limitations

The Impact Score is a market simulation estimate, not a certified benchmark. Distributions are generated from publicly available data and AI-calibrated parameters. Results should be interpreted as indicative positioning within a product category, not as absolute environmental claims.

9. Full LCA: the 16 EF v3.1 (PEF) impact categories

Beyond the carbon footprint, each report can be expanded to a full Life Cycle Assessment that quantifies the 16 impact categories of the Environmental Footprint (EF) v3.1 method, the basis of the European Commission's Product Environmental Footprint (PEF) (Recommendation (EU) 2021/2279). The carbon footprint (climate change) remains the headline figure and does not change when you expand. Full LCA adds the other 15 categories on the same inventory basis.

The 16 categories are: climate change (kg CO₂ eq), ozone depletion (kg CFC-11 eq), human toxicity cancer (CTUh), human toxicity non-cancer (CTUh), particulate matter (disease inc.), ionising radiation (kBq U-235 eq), photochemical ozone formation (kg NMVOC eq), acidification (mol H+ eq), eutrophication terrestrial (mol N eq), eutrophication freshwater (kg P eq), eutrophication marine (kg N eq), freshwater ecotoxicity (CTUe), land use (Pt), water use (m³ depriv.), fossil resource use (MJ), and mineral and metal resource use (kg Sb eq).

The impact vectors for each category are computed directly with Ecoinvent 3.9.1 via Brightway, applying the official EF v3.1 characterisation factors. They are never estimated or scaled from the carbon result.

Missing data and coverage

Coverage is declared per category: the share of the product's impact whose components have characterised data for that category. When a component has no data for a category, it is reported as “n/a” and is not replaced with a guessed value, in line with the ISO 14044 treatment of missing data (§4.2.3.6).

Indicators to interpret with care

The three toxicity categories (human toxicity cancer and non-cancer, and freshwater ecotoxicity) carry higher model uncertainty and are shown in a separate subgroup of additional indicators to be used with care. Water use is computed with the AWARE factor as a global average and is not regionalised, so it does not distinguish arid from humid regions. It is flagged with the same caution.

Full LCA is in beta. The SGS-validated carbon footprint is unaffected by the expansion.

If you are interested in the Life Cycle Assessment of your products, you can request information at devera.ai.

For more information about the calculation methodology, contact info@devera.ai.