View Only

Using AI for Scope 3 emissions data categorization

By Mamatha Venkatesh posted Tue February 06, 2024 09:27 AM


Authors: Mamatha Venkatesh, Georgina Macrae 

Scope 3 emissions take up more than 70% of a business’ total carbon emissions on average.
The GHG protocol corporate standard divides scope 3 emissions into upstream and downstream emissions and then classifies them into 15 distinct categories. These categories rely on getting the right data from suppliers and other parts of the value chain. Disclosing Scope 3 emissions is thereby very complicated. The time to start disclosing Scope 3 emissions is now. 

Scope 3 reporting is becoming mandatory across the world.

Steps to start, and to improve, your scope 3 reporting.

·       Start by complying to scope 3 reporting requirements.

·       Use data of what you spend across your value chain and apply emission factors to it to baseline the scope 3 emissions.

·       Your Purchased Goods and Services are likely to be the biggest portion of your scope 3 emissions.

This ‘Scope 3 Category 1’ may include emissions from hundreds or thousands of suppliers, and where the money is spent likely varies month by month.

The straightforward way of starting to baseline your scope 3 emissions is to find your Enterprise Resource Planning (ERP) system and apply emission factors row-by-row. Emission factors exist that suggest emissions for spend-data. Of these, the Eora global supply chain database is well recognized. 

IBM Envizi uses these emission factors to help businesses to automatically categorize their spend-data, from ERP system data.  Envizi uses emission factors from US EPA Climate Leaders Program, e-GRID USA, Intergovernmental Panel on Climate Change (IPCC), IEA National Electricity Factors, Australian National Greenhouse Accounts, DEFRA (UK), and NZ Ministry for the Environment. Yes, it is indeed tedious.

 Let’s consider a scenario here:

This is a ‘spend-based’ calculation: your system says you spent £850 on paper cups in January 2024, you find the relevant emission factor that says, “for every $1 spent on paper cups, associate x kg of CO2 equivalent” and you multiply the two to baseline your company’s carbon from those cups. The reality of this is that someone filters emission factors row by row through spreadsheets. The spreadsheet will vary month by month, so it could be 1000s of rows of manual processing. This could take days every month and involves human error.

Using IBM Envizi, this painstaking manual process can be avoided and replaced with automation. Envizi uses Natural Language Processing (NLP) to automatically categorize and apply the most appropriate Eora66 emission factors scope 1, Purchased Goods and Services data from ERP systems. NLP is particularly useful for scope 3 spend-based emission calculations because of their scale, and their variability.

This time-saver gives sustainability managers time and space to focus on getting the relevant data for average or other proxy methods of scope 3 calculations, and then getting supplier-specific or site-specific values from those biggest suppliers, to optimize your value chain emissions.

Using Envizi to help to automate your scope 3 base-lining highlights those purchased goods and services which have a higher carbon impact.

This allows the user to focus on getting more accurate data for those areas of spending, or to choose alternatives such as providing ceramic mugs, or washing areas for reusable cups, in offices to reduce the numbers of paper cups that are purchased.

IBM Envizi supports over 100,000 emission factors across scopes 1, 2 and 3 for our clients, to simplify these calculations and to provide better trust and accuracy in sustainability data.

IBM has guidance on how to get started with some of these regulatory standards: Checkout these links for more information on ESG frameworks from IBM .

You can find more details on GRI, SEC, CSRD, ESRS, CDP, BRSR, SASB, SECR, GRESB, TCFD, NGER

Refer: ESG reporting framework with IBM Envizi

Ask an IBMer for more information and try Envizi for yourself!

Authors: Mamatha Venkatesh, Georgina Macrae 





Tue February 13, 2024 05:53 AM

Let us make best-practice GHG calculations pain-free! Thanks for posting, Mamatha, it was great to co-author this with you.