From Messy Data to Material Decisions: How Watershed Builds with ecoinvent

Most companies trying to reduce their emissions face the same early problem: the data they have is the wrong kind. 

Spend-based carbon accounting is a common starting point. It is fast to set up, and it uses data that almost every company already has — procurement spend by supplier. But as Tarik Moussa, Sustainability Advisor at Watershed, puts it:

“If all you have is the amount of money you have spent on a supplier, you’re not able to capture any nuance on the exact type of products you’ve bought from that supplier, which means any decisions you make to shift money towards greener products does not result in a reduction in your Scope 3 measurement.” 

In other words: spend-based data is a reasonable foundation for reporting, but it is less well-suited to the targeted reduction decisions that make a difference.  

The gap between measuring emissions and actually reducing them is the problem Watershed was built to address. 

The Challenge: From High-level Accounting to Product-level Decisions

Watershed is a sustainability AI platform that helps enterprises measure and report their emissions data and, ultimately, reduce them. Its customers are large companies with complex supply chains, many of whom come to the platform with data that is incomplete, inconsistent, or locked in systems that were never designed to answer environmental questions. 

The shift Watershed has made over recent years is from high-level, spend-based approaches toward activity-based, product-level emissions measurement. The challenge, as Moussa explains, is data quality and consistency:

“How do you get good enough data and good enough measurements that you can make business-critical decisions?” 

“The goal is not to wait five years for better data before taking action,” Moussa continues. “We need to worry about perfect data. We need to focus on what decisions we can make on the data we have, what questions we need to ask to get better data, and then how to get to better decisions. The goal is ultimately to take action based on the data that we have today.

“Standardized databases like ecoinvent help companies get real insights and move forward immediately.”

The Solution: AI-powered Mapping to a Trusted Scientific Foundation 

When customers bring Watershed data on their purchases — bill of materials, purchase orders, spend records, even descriptions in mixed languages or internal acronyms—Watershed’s AI agents clean that data and map it to  ecoinvent’s life cycle inventory database. For straightforward inputs like purchased cotton or steel, this mapping is direct. But most procurement data is far more complex. 

“Many of the purchases they’re making are not simple products,” Moussa explains. “They’re complex finished goods, and there’s no neat one-to-one mapping to an existing emissions factor.” 

To generate Product Carbon Footprints (PCFs) for these cases, Watershed builds what it calls a production graph, a digital twin of the supply chain. The AI models each step in the production process, from raw materials through transport, manufacturing, and finished goods. Critically, ecoinvent’s role here goes beyond providing emission factors at the end of that process. The granular, transparent structure of the ecoinvent database—its representation of upstream activities, supply chains, and production systems—is part of the logic the AI uses to infer and construct those production graphs in the first place. ecoinvent is being leveraged both for the breakdown and for the mapping. 

“Using this production graph, we’re then able to map each step of production to existing standardized emissions factor databases like ecoinvent,” says Moussa. “What this means is customers get a full view of what’s driving the emissions from the products they purchase.” 

The production graph is also a living model. As customers gather better supplier data over time, they can refine it to more closely reflect real-world production—making the analysis and resulting PCF progressively more precise without requiring a rebuild from scratch. 

In Practice: Burton Snowboards  

Burton Snowboards is one example of what this approach makes possible. 

Like most manufacturers, Burton carries the majority of its emissions in Scope 3. The challenge for its R&D team was that they had always worked on individual components in isolation, without visibility into the full environmental footprint of a product across its lifecycle. 

Using Watershed, Burton was able to map its purchases to ecoinvent emissions factors and generate PCFs for different product variants. This meant the team could model, on a like-for-like basis, the climate impact of switching specific materials—comparing virgin plastic against recycled TPU against other alternatives—in a way they had not been able to do before. 

The result was not just a climate insight. It was a financial one. Switching to recycled TPU could reduce emissions significantly and potentially save Burton over $130,000 per year across their product portfolio. 

“What was really valuable to them,” says Moussa, “was not just being able to model that data from a climate angle, but being able to understand the financial impact as well.” 

The Bigger Shift: From Reporting to Deciding 

The Burton example points to something Watershed is working toward more broadly. For most sustainability professionals, the daily reality is that the vast majority of their time goes on data collection, cleaning, and reporting—leaving very little time for the strategy and implementation work they set out to do.

Moussa puts it plainly: the industry has historically operated on a 90/10 split, with 90% of effort spent on measurement and only 10% on actual decarbonization. His view is that AI, combined with standardized databases like ecoinvent, can flip that equation—freeing sustainability teams to spend far more of their time on the work that reduces emissions.

“With AI, with standardized databases, we’re looking to flip the script on that,” says Moussa, “so you can quickly get to a view of your hotspots, quickly model out the impact of different decarbonization changes, and then spend ninety percent of your time actually implementing those.” 

For software developers and product teams building in the sustainability space, this is the opportunity: tools that remove the measurement burden and put decision-making intelligence directly in front of the people who need it.

The quality and consistency of the underlying emissions data matter enormously to how useful those outcomes are. It is the reason Watershed built with ecoinvent. 

See Moussa and other professionals in the sustainability field talk more on their work in our panel recording: Environmental Intelligence in Practice: How Leading Companies Turn Data into Business Value.  

To explore building with ecoinvent, visit our partners page.