University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Optimizing the employee scheduling of the service bureau of Company X
Boonman, Floris-Jan (2025) Optimizing the employee scheduling of the service bureau of Company X.
|
PDF
12MB |
| Abstract: | This research was conducted at Company X, a multidisciplinary foot care organization, to improve the efficiency and effectiveness of its central service bureau. The service bureau, which handles both inbound and outbound communication tasks, has experienced increasing workload complexity, primarily due to the company’s expansion. The main operational challenge addressed in this study is the high proportion of missed incoming phone calls, despite adequate staffing levels, due to the lack of data-driven workforce planning. The central research question of this thesis was: “How can a data-driven scheduling approach be used to improve the quality of service for incoming calls at the service bureau from 68.1% to 80%?” To address this, a six-phase research approach was adopted. First, the current situation was assessed, including existing scheduling methods and performance metrics. The Quality of Service (QoS), defined as the percentage of calls answered within 120 seconds, was introduced as a more patient-centered performance indicator. The current QoS stood at 68.1%, with a desired target of 80%. Next, relevant literature was reviewed to identify suitable workforce planning techniques. A hybrid modeling framework was selected, combining queueing theory (Erlang C and Erlang A models) for determining required staffing levels and a Mixed Integer Programming (MIP) model for generating feasible employee schedules. The Erlang C model, despite not accounting for abandonment behavior, provided the best fit to historical call data. In the solution design, historical data from 2025 were used to statistically model call arrivals, service durations, and abandonment rates. These were translated into staffing demands for half-hour intervals using the Erlang models. The resulting demand tables were then used as input for the MIP model, which optimized schedules under real-world constraints such as employee availability and working hour regulations. A detailed Python implementation was developed for both model components. The model was tested using historical data, comparing predicted versus actual performance over several weeks. Results showed that the data-driven approach could generate schedules that meet the 80% QoS target in most intervals. A comparison of model-generated and actual schedules demonstrated increased efficiency and better alignment of staff supply with demand. A sensitivity analysis revealed how changes in personal queue time impact staffing requirements. From a practical perspective, this research provides a scalable and structured framework for employee scheduling that reduces manual workload for planners and better aligns staffing with call demand. It also led to the creation of a performance dashboard, increasing visibility into individual and team performance. This has shifted managerial focus toward the service bureau and enabled data-driven performance monitoring. Insights such as the impact of personal queue time have already prompted managerial action, and the company is now actively exploring the integration of the model into dedicated planning software to support ongoing operational improvements. From a scientific perspective, this research contributes by comparing the performance of Erlang A and Erlang C models in a real-world setting, highlighting the practical implications of incorporating customer abandonment into staffing decisions. Additionally, it introduces the inclusion of personal queue time into the service rate, resulting in a more realistic and applicable queuing model for service environments. These adaptations help bridge the gap between theoretical modeling and operational practice. Key limitations include reliance on simplified assumptions (e.g., identical employee productivity) and use of average input parameters rather than dynamic forecasts. Future research should explore simulation-based models to account for abandonment and variability in employee performance and consider integrating outbound tasks and workload forecasting to improve planning accuracy. |
| Item Type: | Essay (Bachelor) |
| Faculty: | BMS: Behavioural, Management and Social Sciences |
| Subject: | 31 mathematics, 85 business administration, organizational science |
| Programme: | Industrial Engineering and Management BSc (56994) |
| Link to this item: | https://purl.utwente.nl/essays/107910 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page