Platform / Sustainability Index
Module 2 · Sustainability Index
One number that tells you how sustainable a product is
The Index turns materials, logistics, production, packaging and prototyping into a score from 0 to 100. Calculated for every product, from day one, even while some data is still missing.
Example product · hoodie, organic cotton
Air freight lowers the score.
- Materials56/6565%
- Logistics9/2020%
- Production8/1010%
- Packaging3/33%
- Samples2/22%
value predicted by the ML model
Five pillars, configurable weights
The default weights reflect the real share of each category in the footprint of fashion and footwear. For other segments, such as home goods, the weights are set differently, in line with the retailer's strategy.
Materials
65%Origin and production of the raw material, for example virgin polyester, recycled polyester or certified material. The largest effect on water, energy and chemicals, hence the highest weight.
Logistics
20%Carbon footprint of transport from supplier to distribution centre, from distance, mode of transport and product weight.
Production
10%The supplier's environmental certificates for water, energy and waste.
Packaging
3%Packaging material and its end of life: recycling, composting, biodegradation.
Samples
2%A digital sample instead of a physical one earns the maximum score.
Product dashboard
The score and its causes in one view
Every product has a dashboard with the overall score and the points in each pillar. Buyers see strengths and weaknesses at once, for example a certified material and a score pulled down by air freight.
- List of products that need attention because of a low Index
- Score details in every category
- Similar products ranked by Index, with purchase price, retail price and expected profitability
- Index recalculated automatically with every change to the product, supplier or order
- Repeatable: the same data always gives the same score, tested on 500 products in 10 runs
Management view
The Index as a KPI
A score for each product, not for the whole company as in indices like the Higg MSI. That makes the Index usable as a target for buying teams, a basis for comparing suppliers and categories, and a data source for non-financial reporting.
Set and track sustainability targets for the collection.
Product data structured for reports, including in Power BI.
Machine learning
An Index from day one, not after weeks
Early in the process, data on materials or transport is missing, and collecting it from suppliers takes weeks. An ML model predicts the missing attributes from the data already there: brand, colour, purchase price, product hierarchy, supplier, country of origin, season and product type.
Every predicted value is marked in the interface. When the supplier provides the real value, the Index recalculates automatically. The Index stops being an after-the-fact report and becomes a design criterion, on a par with price and quality.
F1 for predicted attributes, target at least 70%.
| Lining | 83.3% |
| Outsole | 84.4% |
| Insole | 84.3% |
| Upper | about 79% |