Operationalising software carbon intensity through key performance indicators : an industrial thesis

Abstract

The growing environmental impact of digital technologies has intensified the need for robust sustainability metrics in software engineering. The Software Carbon Intensity (SCI) metric, introduced by the Green Software Foundation, quantifies the carbon emissions attributable to software by accounting for both operational and embodied emissions per unit of useful output. However, SCI is primarily used as a standalone assessment score without integration into organisational contexts to drive change in business operations. This thesis investigates how SCI can be integrated into organisational Key Performance Indicators (KPIs) to enable continuous monitoring and improvement of software sustainability. The research is conducted in collaboration with a major financial institution in the Netherlands and employs a mixed-methods approach, combining stakeholder input with quantitative SCI data analysis from a production-grade cloud application. The study details the challenges that come with calculating SCI in a real-world setting, from data granularity to functional unit selection, and demonstrates a practical method for implementing an SCI-based KPI in an enterprise context. Results show that while such KPIs are feasible and valuable for strategic sustainability management, their effectiveness depends on the availability of transparent emissions data and careful integration within software engineering processes. This work advances the practical use of software sustainability metrics and guides practitioners seeking to bridge the gap between environmental objectives and digital infrastructure management.