How Reformation tracks the footprint of every product with RefScale
Reformation measures the carbon and water footprint of each garment with an internal life cycle tool. This case study explains what RefScale covers, how it is verified and where AI does and does not fit.
KEY TAKEAWAYS Summary by the editors
- Reformation launched RefScale on 22 April 2015 to show the water and carbon dioxide used to make each garment, compared with an industry standard for a similar style.
- RefScale is an internal life cycle assessment tool that covers nine stages, from fabric manufacturing through customer garment care to end-of-life, and reports carbon and water results on product pages.
- Point B's independent assurance for calendar year 2023 tested sample purchase orders and styles, and found the carbon and water methodologies appropriate, while recommending supplier-specific emission factors where possible.
- Reformation does not describe RefScale as an artificial intelligence system; it is a data-driven calculation model, and the sources reviewed name no machine learning component.
- Point B's engagement covered the tools' methodology and data, not their functionality, and the assurance report does not publish the underlying CO2e or water figures.
Reformation, the US fashion brand, tracks the environmental footprint of its products with RefScale, an internal life cycle assessment tool that estimates carbon and water impact for each garment and shows it on product pages. The company has not described RefScale as artificial intelligence. It is better understood as a structured data model, and the case shows why that foundation matters before any machine learning is added.
What is RefScale?
Reformation launched RefScale on 22 April 2015, for Earth Day. It shows shoppers the amount of water and carbon dioxide used to make each piece, set against the industry standard for a similar style. The company describes it today as an internal life cycle assessment tool that estimates the carbon and water footprint of products and compares each one with the average of clothes bought in the United States.
What does RefScale measure across the product life cycle?
According to Reformation, RefScale covers most processes in a garment's life, from fabric manufacturing to end-of-life. Water is weighted by scarcity and region. The company lists these stages:
- Fabric manufacturing
- Fabric dyeing
- Material transit
- Product manufacturing
- Commercial garment wash
- Packaging
- Shipment
- Customer garment care
- Garment end-of-life
Inputs are water, energy and raw materials, and the outputs are CO2 emissions and water. Reformation publishes its methodology as a PDF, shows results on each product page and reports totals in its sustainability report.
How is the data checked?
Reformation says it works with a third-party consultant each year to review and verify RefScale. For calendar year 2023, Point B issued an independent assurance statement dated 26 March 2024. It covered two tools, one for apparel and one for shoes and accessories, across worldwide manufacturing and distribution, for greenhouse gas emissions and water use.
Point B reviewed methodology documents, emission factor sources and data collection methods, including how product weights are calculated. It sampled ten random purchase orders across ten apparel product classes, twelve shoe styles and nine bag styles. Emission factors were spot-checked against Higg MSI updates. Point B concluded that the tools sufficiently represent the relevant life cycle stages and that the carbon and water methodologies are appropriate.
| Aspect | Covered | Not covered or caveat |
|---|---|---|
| Scope | Apparel and shoe and accessory tools, cradle to end-of-life | The statement is intended for Reformation and for Change Climate certification |
| Method | Assumptions, limitations, emission factor sources | Not assurance of tool functionality |
| Data | Sampled purchase orders and styles, unit conversions | Findings on primary data sources; supplier-specific factors recommended |
| Outputs | Carbon and water results as used in the corporate GHG inventory | Actual CO2e and water figures are not in the statement |
Where does AI fit in sustainability data?
The sources reviewed do not say that Reformation uses machine learning for RefScale, and this article does not claim it. The relevance for AI is the data prerequisite. Models that estimate footprints, flag missing supplier data or answer customer questions can only be as good as the product-level data behind them. Point B's main finding was about primary data sources: it suggested supplier-specific factors where applicable and reliable secondary factors for footwear and apparel.
- Product-level footprints need weights, materials and processes recorded per style.
- Emission factors must be sourced, versioned and updated.
- Independent verification supports credibility, and does not replace it with certainty.
Why does footprint data matter for AI and regulation?
Product-level environmental data has become more important as fashion companies prepare for digital product passports and for stricter rules on environmental claims. A brand that already holds per-style material, weight and process data can answer such requests faster than one that must reconstruct them. The sources reviewed do not discuss regulation in connection with RefScale, so this is an inference about where such data is useful, not a statement about Reformation's plans.
AI tools could help with narrow tasks around such a system: extracting material composition from supplier documents, flagging outliers in submitted data, or drafting product-page explanations. Each of these still depends on a verified calculation engine underneath, and each introduces a risk of confident error. A language model that explains a footprint wrongly is worse than no explanation.
| Prerequisite | Why it matters | Evidence in this case |
|---|---|---|
| Per-style bill of materials | Footprints depend on fibre type and weight | Point B checked how product weights are calculated |
| Sourced emission factors | Results depend on factor quality | Spot-checked against Higg MSI updates |
| Supplier-specific data | Generic factors hide real differences | Point B suggested supplier-specific factors where applicable |
| Published methodology | Lets outsiders assess assumptions | Methodology published as a PDF |
| Independent review | Builds trust in the numbers | Annual third-party review, with a 2023 assurance statement |
A final caution: comparing one brand's footprint with an average of US clothing purchases depends on the assumptions of the comparison, which only the published methodology can show.
It is also worth noting what is not in the public sources. They do not state how many products RefScale covers, what share of Reformation's footprint comes from each life cycle stage, or how results have changed over time. The assurance statement is explicit that it does not publish the underlying figures and that it is intended for Reformation and for a certification submission. A reader should therefore treat RefScale as a documented method with independent review of its approach, not as proof of any particular environmental outcome.
For editors and analysts, a useful habit is to ask three questions of any footprint claim: what is the boundary, which factors were used, and who checked them. RefScale's published materials answer the first two in outline and the third through annual third-party review, which is more than many brands disclose.
What can other brands learn from this?
The case shows three practical points. Per-product footprint data is a long-running data engineering effort, started by Reformation in 2015. Verification is an annual process, and its findings, including recommendations to use supplier-specific data, remain open work. And publishing the methodology lets outsiders assess the assumptions. Brands planning AI-based traceability or product passport work should first check whether they hold this kind of structured, verified product data.
Readers should note that some figures on Reformation's sustainability pages could not be read reliably during research, so this article cites no targets or progress percentages from them.
Frequently asked questions
What is RefScale?
RefScale is Reformation's internal life cycle assessment tool. It estimates carbon and water footprint for each product, compares it with the average of clothing bought in the United States, and shows the result on product pages.
Does Reformation use AI to measure its footprint?
The sources reviewed do not describe RefScale as artificial intelligence. They describe it as a life cycle assessment tool with published methodology and annual third-party review.
Is RefScale independently verified?
Reformation says a third-party consultant reviews RefScale each year. Point B issued an assurance statement for calendar year 2023 on the tools' methodology and data, and did not assure their functionality.
What does RefScale cover?
It covers nine life cycle stages from fabric manufacturing to garment end-of-life, with carbon and water as the main outputs. Water is weighted by scarcity and region.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
