When I started my internship at Catawiki, I took up the role of an assistant at the Customer Relationship Management (CRM) team. My daily responsibility was to make marketing marketing emails, in-app notification and mobile push based on the design provided by the Studio team in Catawiki. As I gradually became more familiar with everything in the CRM team, I was also invited to the design phase and analysis phase of a marketing campaign. Most of the campaigns at Catawiki involved a larger number of audience, which enables A/B testing for analyzing campaign effect. The difference between the campaign group and control group in an A/B testing was interpreted as the effect of the corresponding marketing campaign. Nonetheless, in some campaigns targeting high value buyers and sellers of Catawiki’s platform, the audience size was not large enough for A/B testing and all the target audience had to fall in the campaign group. With the collaboration between Marketing department and Data department, how to measure the marketing effect in this particular case was still under exploration in Catawiki. Given that my Communication Science program had some statistics background, I was encouraged to explore this topic as a personal project and propose a potential solution with the help from the Data team. This project has been going on for 2-3 months and I am pleased to share some personal gains from this experience from the point of view of a Communication Science background student.
Say ‘Hi’ to statistics again
The course of MCRS or SMCR might not sound very appealing to many Communication Science students. It is imaginable that some of us might be happy about the thought of putting all the required statistics aside once they bid farewell to academia and embrace the real world. Some others might feel that our statistics knowledge is not competent enough. Well, from my personal project experience at Catawiki, I would say that both thinkings are not fully correct in the case of marketing. Firstly, although statistics is not strictly needed in the design and operational phase of a marketing campaign, it is the most important tools when analyzing the result, which is an indispensable step for any marketing actions. Secondly, all we have learnt in the MCRS and SMCR courses are useful and many of them are directly applicable, but more often than not they serves as the basics. I remember the proudness I felt when I first mentioned dependent sample t-test as the solution to the colleagues in the data team and it received an immediate positive comment. However, t-test marked only the beginning. After t-test, we have later tried Wilcoxon test, permutation and Bayesian estimation on the same dataset and compared the difference between them. So for every Communication Science student who would like to or is required to involve in the process of result analysis in marketing, on the one hand, please be confident about your statistics knowledge! They are useful and most likely sufficient for a good start! On the other hand, please be prepared to learn more and go wider in not only statistics, but also python, SQL or R (we don’t have Qualtrics or SPSS anymore).
No ideas? Ask AI!
The launch of Chatgpt in 2022 has changed our world drastically. Regardless of all the controversies and discussion around it, it offers you a super powerful assistant that is always available. In the real business world, the situations are so versatile that one could always feel knowledge insufficiency. This feeling is especially prominent for new employees who lacks experience in the corresponding field. In this case, AI would be an amazing tool to help bridge the knowledge and experience gap. Using answers from AI to gain inspiration is commonplace even among the senior employees, so feel free to use it for your own project when the company’s policy permits. When I was working on this personal, data-related project at Catawiki, the help from Chatgpt was indispensable. I needed its help to generate SQL code to retrieval the relevant data from the data warehouse for analysis. I needed its help for alternative statistical test when most of the tests I knew did not suffice the real situation (I know Wilcoxon test as an alternative test to t-test only thanks to AI). I also needed it to help me translate the statistical tests into python code. As you can see, AI is really useful for our work and using it does not hinder your qualification for the position. Moreover, you can really learn a lot in this process with the help of AI and gradually you would be able to use it less and less.
Issue ownership: Yours are yours!
Project ownership is one thing that most students take time to adjust to when they enter into the real working world. With different methods in hand, it is your responsibility to pick one that you prefer the most and present it as the final solution to your own project. This mindset is more difficult to adjust to under the circumstances where you are actually not sure which solution would be the best. During my personal project, I remember the moment when I put all the different statistical tests on the table during my meeting with a colleague from the data team, hoping he could make a decision on which one would be the best. His answer to me was “It is your personal project. You pick the one you believe it’s the best and we will use that one to test-analyzing the next campaign result.” It is at this very moment do I have a vivid impression of the responsibility of a personal project. But please don’t get intimidated by this. No one is expecting you, an intern and a beginner, to solve a problem at company-wide level. After careful work, you have everything to present your result clearly and confidently and what you have done could serve as a building brick to the final solution for the company. Overall, as a beginner, the curiosity, courage, potentiality and hard work you have devoted to your personal project is valued the most by the organization.