Marriott International Implements Data Automation Tools & Artificial Intelligence for Scope 3 Emissions Reporting Enhancements

Commercial, Energy Efficiency, GHG Emissions  -  November 15, 2024 - By Better Buildings, U.S. Department of Energy

Marriott International Implements Data Automation Tools & Artificial Intelligence for Scope 3 Emissions Reporting Enhancements

Overview: As scope 3 emissions reporting becomes of increased interest, Marriott International (Marriott) decided to improve its data management practices by implementing an AI solution. Marriott’s data practices for scope 3 emissions focus on spend data; Marriott takes inventory from its General Ledger spend (among other sources for added detail) and assigns emission factors based on the nature of the spend. Previously, Marriott used Excel for this process but found the process to be time-intensive and lacked accuracy given the diversity of descriptions associated with spend data. To increase efficiency and accuracy, Mariott automated this process using Artificial Intelligence (AI), resulting in a 5-10% reduction in scope 3 emissions and a decrease in the number of labor hours required. 

Barrier: Marriott’s method for calculating scope 3 emissions required manual entry, was time-intensive, and lacked accuracy. 

Solution: Acquired an AI product and trained it using previously validated emission factor mapping logic for Marriott’s spend data. 

Outcome: The AI product increased accuracy in emissions reporting due to the integration of new EPA-published emissions factors and saved an estimated 250 work hours.

Process: The first step in updating Marriott’s data automation process was to create an Inventory Management Plan (IMP). Marriott contracted a carbon accounting consultant to draft the IMP but has since taken over the task of updating the IMP. This created the foundation of Marriott’s scope 3 emissions reporting. Moreover, Marriott has an IMP for all emissions data and calculations to track carbon accounting processes over time. They will mature and the IMP will record the history of changes and improvements.

The next step was to integrate the AI process of mapping emissions factors which required collaboration with the AI developer and the development of an internal Climate Action Program (CAP) database. Marriott’s data practices for scope 3 emissions focus on spend data. This means Marriott takes inventory from its General Ledger spend (among other sources for added detail) and assigns emission factors based on the nature of the spend. Given the diversity of descriptions associated with the spend, it is difficult to efficiently organize all spend by emission factors. This is where it became necessary to refactor the CAP database to report spend at a hotel/business entity level with whatever detail is provided and send it to the AI tool for fast processing. Between the speed of automation that the AI brought and a refactoring of the CAP database to remove any methodology assumptions and enhance data granularity, Marriott was able to feed the AI platform the enhanced CAP data and was then able to calculate emissions.

The integration process took approximately one year to complete. The organization had to audit its methodology, re-curate data to satisfy data ingestion purposes for the AI model, and ultimately train the model for emissions factor mapping. Additionally, due to the diversity of Marriott’s data, setting up the CAP database took a significant amount of time. However, future reporting periods will see increased automation due to the platform's ability to intelligently recognize and map spending otherwise unassociated with emissions factors.

A substantial benefit of the AI tool is that it is not only designed for reporting spend-based emissions but also for supplier-based reporting which is consistent with GHG protocol and noted as having the highest accuracy within its framework. This means the AI platform will support Marriott’s Procurement Team in a supplier/vendor engagement campaign to source emissions factors directly from suppliers. This work is ongoing and will continue over the foreseeable future. Data obtained by suppliers will be joined to the spend data with the original spend data removed to avoid double counting as emissions reporting accuracy grows with continued supplier engagement.  

Outreach: Gaining internal support of the AI approach to data management was crucial to its success. To gain support, the Marriott data team met with internal stakeholders, specifically internal finance business partners, for several months. These meetings included reviewing data, emissions mapping, and emissions calculations. 

Measuring Success: Marriott expects to save on costs over time due to the decreased labor required to calculate and track scope 3 emissions data. Marriott also expects a slightly accelerated timeline in scope 3 decarbonization by keeping emissions factors as up to date as possible and incorporating supplier-based emissions factors. 

Outcomes: By incorporating a more recent emissions factors set published by the EPA, the AI emissions measurement methodology allowed Marriott to reduce its scope 3 emissions by 5-10%. This also reflects the most recent updates to grid electrifications. 

Marriott is also able to estimate time saved by comparing workloads for measuring emissions from prior years. Current estimates mark 250 hours saved and this number will grow each year. Moreover, the CAP database has enabled better data sharing across the organization and lends itself to easy-to-write queries for answering ad-hoc questions from stakeholders and hotel engineers. 

 

This column originally appeared on the Better Buildings website.

The Better Buildings Solution Center houses over 3,000 resources shared by Better Buildings partners and other stakeholders. These replicable solutions help organizations bolster their bottom line, advance technology innovation, create jobs, and spur energy efficiency investments.

 


« Back to Energy Management

  • LinkedIn
  • Subscribe

Smart Energy Decisions Content Partners