Exploratory Data Analysis of Cloud Service Tickets

Context:

A medium-sized cloud service provider in Switzerland must be able to successfully manage its customer service processes in order to maintain customer satisfaction and the company's profitability. Despite significant investment in digital solutions, the company faces ongoing challenges in efficiently managing and resolving customer queries. This study examines these operational inefficiencies, focusing on the processing of support tickets within the ServiceNow platform.

Goal:

The study aims to determine how the company can improve its operational efficiency and customer satisfaction by eliminating inefficiencies in customer service ticket management processes. The study also aims to identify potential improvements in data structure and processing to ensure that processes can be monitored in the future.

Methods:

To achieve this goal, a comprehensive data cleaning and pre-processing methodology was used in the study to ensure accuracy and consistency. Advanced text mining techniques such as Term Frequency-Inverse Document Frequency (TF-IDF) and Latent Dirichlet Allocation (LDA) were used to structure and analyze the unstructured text data. In addition, clustering techniques such as K-Means, DBSCAN and hierarchical clustering were applied and compared to divide the data into meaningful groups.

Results:

The analysis revealed various clusters with different processing times, which revealed important inefficiencies. The analysis revealed that issues related to "Support Infrastructure" and "Network and Server" topics were prevalent in the clusters with longer processing times. The results indicate that requests often remain in "Open" or "On hold" status due to delays in initiating a resolution. In addition, business service and urgency information was often insufficient for effective analysis, indicating the need for better structuring and prioritization.

To address these inefficiencies, the company is recommended to improve the visibility and management of tickets in "open" and "on hold" status, improve the structuring of business service information, and provide additional training for handling specific issues. Future research should focus on the development of predictive models to further optimize the support process and increase customer satisfaction.