Data work, automated.

Carbonfact builds customized python connectors for each client to automate data collection from their systems. The platform can retrieve information from any ERP (e.g., SAP, Oracle, Microsoft Dynamics), PLM (e.g., Centric, Lectra, PTC), traceability solution, or structured files (.csv, .xls). The process typically begins with static exports, then moves to live connections, and requires minimal effort from the client. Scope 1 and 2 data can be ingested one by one in the platform, with guided templates, or through our AI Import studio.

Yes, Carbonfact is specifically engineered for scalability, making it capable of handling large-scale operations. The platform has been successfully implemented by major global brands, including high-revenue companies such as On. Carbonfact's architecture allows it to efficiently process:

• Tens of thousands of product references
• Tens of millions of purchase orders annually
• Detailed SKU-level data for comprehensive analysis

A key example of Carbonfact's large-scale data ingestion capabilities is its ability to process Purchase Order data, enabling precise mapping of quantities at the SKU level. This granular approach offers two significant advantages:

• It allows brands to prioritize their data collection efforts based on actual production volumes.
• It enables the reconciliation of product-level changes with overall company sustainability trajectories.

This scalability ensures that as fashion brands grow and their data needs expand, Carbonfact can continue to provide accurate, comprehensive sustainability insights without compromising on performance or detail.

In three ways:

• Via an API connection
• Via CSV uploads
• Via suppliers.carbonfact.com
• Through the platform itself. Due to the amount of customers and suppliers already in our database, a portion of your suppliers likely already have data in our platform.

Yes, Carbonfact's connector includes automated tests to identify missing and potentially erroneous data. Users can view incomplete or suspicious products directly on the platform or via a weekly summary newsletter. The platform also uses a "Carbon Uncertainty" metric to indicate the accuracy of results. We define the metric Carbon Uncertainty as the difference between the maximum and minimum emission factor value of a material, multiplied by its volume. This provides clear direction for clients to prioritize which missing data should be augmented first.

Carbonfact includes a normalization system to clean "dirty data". It leverages machine learning, trained on anonymized, aggregated data from 200+ fashion brands and suppliers. This system helps detect errors in the data set.

In addition to standard Data Quality Ratio (DQR), Carbonfact has developed a proprietary metric called Carbon Uncertainty. Example: a T-shirt's footprint could be 7.0±1.4kgCO2e (that is a ±20% Carbon Uncertainty). This metric assesses the share of primary versus secondary data and adds a percentage of accuracy to the footprint results, helping brands prioritize their data collection efforts.

Carbonfact relies on two core features to fill data gaps:

1. Carbonfact heuristics: This is a database of product descriptive data (e.g., weights of components, fabric formation techniques) built from the experience of working with 150+ brands and manufacturers. This aggregated and anonymized data is used to fill in data gaps.
2. 2.Rules: Based on the client's understanding of their products, materials, and processes, customers can build specific rules to enrich different segments of products, materials, or suppliers within the platform.

Additionally, Carbonfact uses recognized databases such as Ecoinvent 3.10, EF 3.1 and Ademe (Base Empreinte®) to supplement data. When encountering data gaps, our engine can leverage these databases along with our proprietary technology. For materials or processes where standard databases don't provide sufficient information, Carbonfact's in-house Science team can create custom emission factors based on the latest scientific articles and reports.

Yes, Carbonfact's platform allows full calculation of Scopes 1, 2, and 3. The platform enables activity-based measurement for Scopes 1 and 2, and for several Scope 3 categories including Purchased Goods (3.1, usually 90% of a brand's emissions), Upstream Transportation and Distribution (3.4), Use of Sold Products (3.11), and End of Life Treatment of Sold Products (3.12). For other Scope 3 categories, the platform can use spend-based measurements.

Carbonfact uses a Version Control feature that tracks all calculations and methodological changes. A new version can be released when data is updated, or several versions can be batched and then released together, depending on your desired granularity. This approach allows for easy auditing. The platform's engine and procedures are peer-reviewed annually by PwC to ensure compliance with ISO 14040, PEFCR, and GHG Protocol standards. In Carbonfact, transparency is key to decarbonization. That's why all data (Environmental Factors + your data) is available at all times, just as its sources. Typically, we see customers use their platform on a weekly or monthly basis, with some of the most advanced customers adjusting things like product materials or suppliers in order to lower emissions.

• For the Carbon Accounting solution: First GHG Protocol Compliant report with representative products extrapolated to production volume: 6 weeks
• For the Product LCA solution: 5 weeks to build the connector, 4 months to refine data in full granularity.

Carbonfact's software is deployed to heavily scalable solutions provided by industry leaders (Google Cloud Platform, Vercel, Salesforce) to ensure long-term scalability. When necessary, we perform load-tests before releasing significant architecture changes to production. We can disclose aggregated numbers which prove our large scale data-volume processing capabilities:

Carbonfact's database handles:

- Over 1.6bln Purchase Orders (PO) in 2024

- 800.000 Bill of Materials (BoM) per year

- 1600+ factories

These numbers grow roughly 250% Year on Year.