Recent studies show that software systems account for approximately 6% of global energy consumption. Coupled with new regulations urging French organizations to adopt energy-efficient practices, this has increased the need for energy-conscious development methods. However, a key challenge remains the lack of detailed data on the impact of design solutions on software energy consumption, complicating efforts for software architects who must balance performance and energy efficiency. Without clear data on how design decisions—such as architectural styles, frameworks, and serialization methods—affect energy use, optimizing both performance and sustainability is difficult. Most existing studies on software energy consumption focus on small-scale projects and technologies irrelevant to Berger-Levrault’s environment, where this study was conducted. To address this, we propose a methodology to evaluate the energy consumption of alternative design solutions using the Dynatrace Carbon Impact module, chosen for its compatibility with virtualized and cloud-based environments, and NeoLoad performance testing, selected for its compatibility with Berger-Levrault’s specific technologies. Using this approach, we compared two versions of the same application: a legacy version built with Citrix, Java Swing, and RPC, and a modern version using Angular, Spring Boot, and REST. We also evaluated Protobuf versus JSON serialization. Our findings show that adopting Angular for the frontend and Spring Boot for the backend reduces overall energy consumption by approximately 70%. The Angular frontend uses 30 times less energy than Java Swing. Furthermore, Protobuf serialization reduces energy usage by 33% compared to JSON, due to its faster processing and lower memory requirements. These results highlight the energy-saving potential of modern web technologies and optimized data formats for sustainable software development.