Microsoft Copilot Challenges : User Guide Development to Overcome Challenges in Microsoft Excel

Sammanfattning

This thesis investigates challenges users face when integrating Copilot for Microsoft 365 into their workflows within Microsoft Excel and proposes strategies to address them through effective prompt formulation techniques. The research emphasizes understanding these challenges to improve user efficiency and productivity in Excel. The main problem addressed is the difficulty users encounter in utilizing Copilot for Microsoft 365 within Excel. It aims to identify common issues and propose strategies to overcome them through prompt engineering. By improving users’ ability to prompt effectively, their Excel workflow can be enhanced. The research methodology combines structured questionnaire interviews, distributed on various online platforms such as Facebook, Reddit, LinkedIn, targeting users experienced with Copilot for Microsoft 365 in Excel. Semi-structured, in-person interviews were conducted with four respondents experienced in using Copilot for Microsoft 365 in Excel. The analysis synthesizes results from both methods and previous research to identify challenges, propose effective prompt formulation techniques, and create a user guide. Common challenges include communication issues, data formatting, task complexity, and response quality. Strategies like zero-shot, few-shot, and Chain-of-Thought prompting, along with prompt design practices, are suggested to enhance user interactions with Copilot. The developed user guide aims to assist users in formulating effective prompts, emphasizing clarity and conciseness. This work contributes insights into users’ experiences with Copilot for Microsoft 365 in Excel and offers practical strategies for enhancing user interactions. The findings have implications for both practical and theoretical applications, as they can improve users' efficiency and productivity in Excel. However, due to the small sample size, generalization of these findings should be approached with caution. Future directions can explore advanced prompt formulation techniques and organizational implementations of Copilot to enhance its real-world usage while addressing the limitations of small sample sizes.