Dear Lydia, congratulations on your graduation and your excellent bachelor's thesis! How did you come up with the topic for your thesis? Was it your own initiative or did you develop it together with Tchibo?
After Prof. Dr Andy Witt's first lectures on data science and artificial intelligence, I knew immediately that I wanted to write my bachelor's thesis in this field. At the same time, my first practical assignment in the loyalty department at Tchibo gave me my first taste of customer loyalty, a topic that fascinated me. I wanted to combine these two topics in my thesis.
I then approached the specialist department at Tchibo with a topic proposal in the area of customer behaviour forecasting, and it turned out that the team was currently interested in customer planning issues. In close consultation and depending on data availability, we finally agreed on the topic of forecasting customer lifetime value. This key figure is particularly relevant in CRM, as it helps to target marketing budgets more effectively and better understand customers in the long term. This resulted in a topic that perfectly combines my personal interests with practical benefits for Tchibo.
You changed your track specifically for the minor in data science, so this specialisation was very important to you. When did you realise that you wanted to specialise in data science and artificial intelligence? How did that come about?
During my practical phases, I quickly developed an enthusiasm for data and was fascinated by the derivation of real recommendations for action. I had my first contact with the data science team in the loyalty sector in particular, and the relevance of gaining insights from the database for the development of business strategy and data-driven models further sparked my interest in this field.
At the same time, my interest in data and its potential deepened during lectures at HSBA. The topicality of artificial intelligence and the fact that it is changing almost all areas of life within a very short time have further motivated me to really understand the mechanisms and functioning behind it. The track change to the Data Science Minor was therefore a conscious step for me; I wanted to acquire the theoretical and methodological foundation for data science and machine learning.
You wrote an ‘outstanding’ bachelor's thesis, as your former Professor Andy Witt put it. As far as I understand, however, one of the findings of your thesis is that AI cannot (yet) do everything. Can you explain that to us briefly?
That's exactly what made the topic and the time spent working on it so exciting for me. Right at the beginning, it became clear that Tchibo's data base posed a major challenge for the development of a prediction model. This is due to the complexity of the business model: in addition to the coffee range, the non-food range changes weekly according to the principle of ‘a new world every week’, and sales are spread across three distribution channels. As a result, the data is highly fragmented and forms a difficult basis for machine learning models.
A central principle in artificial intelligence is: ‘Garbage in, garbage out.’ A model can only be as good as the data it is trained with. Therefore, I first had to intensively examine the question of which data needed to be prepared in what form in order to enable a meaningful forecast. In several iterations, I tested different model approaches, starting with classic models such as simple neural networks, which were not suitable for the available data.
The focus of my work therefore shifted from the goal of delivering the ‘ideal prediction’ to the question: How can the use case be modelled at all? Which customer characteristics are useful and which models and methods can still deliver usable forecasts under the given conditions?
Once again, it became clear that not every use case can be easily solved by AI. Rather, performance depends heavily on the specific use case, the available data and the model selection. It is precisely this insight that is valuable – both scientifically, as it highlights the limitations of the methods, and in practice, in order to develop realistic expectations for the use of AI in customer management.
Does Tchibo still derive concrete added value from your bachelor's thesis? And if so, can you give us an example?
The added value for Tchibo lies on several levels. On the one hand, I was able to clearly identify which approaches and methods actually deliver benefits and which are unsuitable. This insight is particularly important for the efficient and joint further development of the model with the relevant stakeholders. This transforms it from a research and testing approach into an operational tool for customer management.
Given that customer behaviour in the FMCG sector is inherently highly variable, the model developed has been shown to achieve remarkable accuracy for certain use cases. Taking into account the inevitable inaccuracies, the results can be used for strategic marketing decisions.
Another contribution of my work was the identification of the decisive features, i.e. the factors that are particularly strongly related to customer loyalty and churn risks. This is central to CRM and the guideline of reducing churn rates and retaining customers in the long term.
In concrete terms, the calculated CLV can be used, for example, to segment customer groups. This allows customers with high sales potential to be identified for targeted marketing measures or, conversely, those for whom further investment is unlikely to be highly profitable. The model thus forms a component for data-based and future-oriented customer planning.
Did your studies prepare you well for working life overall?
From the outset, the dual study programme emphasised the importance of combining theory and practice, which meant that the transition to working life at the end of my bachelor's degree was not a big change. The lectures at HSBA gave me a sound foundation in business administration and methodology, which I was able to apply directly at my practical partner. Even my first practical experiences, for example in the area of loyalty, helped me to identify my professional interests at an early stage and to focus my studies accordingly. This connection was further strengthened by practical projects and academic work, many of which could be directly applied to issues at the partner company, not only in my bachelor's thesis. This allows you to expand your specialist knowledge on the one hand and learn to quickly familiarise yourself with new topics and departments on the other. In addition, working with fellow students from different industries during my studies and with different departments in the company, such as marketing, data science and more, was particularly valuable and highlighted the importance of incorporating different perspectives and developing joint solutions.
In addition to the close integration of theory and practice, I particularly liked the great variety and flexibility of the programme. After choosing finance as my major in the Business Administration programme, I also had the opportunity to choose Data Science as a minor from the Business Informatics branch. I also found the selection of elective courses alongside the core curriculum to be of great added value. This allowed me to pursue my interests very freely and put together an individual study profile.
I also have fond memories of my semester abroad. HSBA offers a wide range of partner universities, and I wouldn't want to have missed this experience. I gained many new impressions, met great people who became lasting friends, and gained valuable experience. Overall, the programme not only advanced my professional development, but also shaped me personally.
