Welcome to our insightful and data-driven blog post, where we uncover the fascinating journey of Marjolein Kok and Sylvie Toonen, two master students on a mission to provide valuable recommendations for FitCapacity's Marketing & Sales department. In this post, we'll take you behind the scenes as they navigate through a rich dataset provided by Microsoft, employing advanced data analysis techniques and machine learning models to unlock market insights.
Join us as we delve into the world of customer demographics, sales strategies, and territory dynamics, gaining a deep understanding of FitCapacity's position in the market. But it doesn't stop there – we'll also explore the ethical considerations and responsible data usage that guided their project, ensuring a comprehensive and socially conscious approach.
Get ready to immerse yourself in the power of data analysis and discover actionable insights that can shape FitCapacity's marketing strategies. Whether you're a data enthusiast, marketing professional, or simply curious about the intersection of analytics and business, this blog post is for you. Let's dive in and explore the exciting results of this compelling project.
Meet the team:
Marjolein Kok (Master Student) - LinkedIn profile
Sylvie Toonen (Master Student) - LinkedIn profile
Background
We welcome you to follow our journey for providing data-driven recommendations for FitCapacity’s Marketing & Sales department. In this blog post, we dive deep into a dataset provided by Microsoft, combining data analysis techniques to create market insights. We navigate through customer demographics, sales strategies, and territory dynamics to gain an understanding of FitCapacity's position in the market. We also take ethical considerations into account by highlighting responsible data usage and discussing the impact this had on our project.
Project overview
We analyzed a dataset given by Microsoft to come to feasible recommendations for the company FitCapacity, and in our case, specifically for the Marketing & Sales department of the company. This allowed us to have a critical look at the market position of FitCapacity, which is essential since the market share of competitors is rising. To come to these recommendations, we combined descriptive analysis by visualizing the given data in Power BI and predictive analysis by creating two Machine Learning models to eventually create one dashboard with our insights related to customer and sales information, which is the most useful data for our chosen department of the company. During this analysis process in Power BI, we tried to consider ethical implications, also in comparison to what we were taught during the course. The diagram in figure 1 displays the various steps that were taken throughout the project.
Figure 1. Steps taken during project
Project journey
To be able to analyze the data for FitCapacity and make useful recommendations, the right data had to be selected and prepared for modeling. We wanted to focus on one specific department to be able to make grounded recommendations and give rich suggestions instead of giving a lot of surface level recommendations on multiple topics. The insights of the Marketing & Sales department, also open doors to data analyses for the other departments. This immediately brought up ethical limitations. What does it mean for our analysis if we choose certain tables, and exclude other tables? Do certain classifications and standards play a role in these tables which have consequences for the representation of the people in the data? Do we maybe have our own biases when choosing certain tables? Using the data schema and dictionary, we looked at the various categories of data and the connecting tables to see which data seemed most useful. The next step was to go through the tables in Power BI, without importing them, to have a first glance at the specific data in the tables. Here we discovered that some tables we selected did not have very useful data for the Marketing & Sales department. we made a list in which we stated what columns seemed most interesting to use and what insights the columns could bring for the Marketing & Sales Department, to eventually have a plan on what data to first focus on.
Technical details
After the data selection, the selected data was loaded into Power BI. We decided to first explore the data within Power BI before conducting Machine Learning. This decision was made based on the fact that we wanted to get more grip on the data and to see what insights the current data gave, before conducting any predictive analyses. Since we did not know where our data came from and what our data exactly meant, we found it crucial to create some insights in our data in Power BI, to get an idea of the environment we are building Machine Learning models for. This made us actually understand the data instead of just mindlessly conducting analyses and it reduced the chances of any bias playing a role in our data as we discussed the insights with each other. As a result, we were able to create correct relationships, clean the data to our preferences, create accurate hierarchies and finally visualize the data in a correct and insightful manner. We created some relationships between tables to help us prepare the data. The next step was to model the data and to perform various analyses. After creating visuals in Power BI, the attention was switched to Machine Learning. From the visuals of Power BI, various predictive dataflows and models of the tables were created. Eventually, two Machine Learning models were created. These two models focus on customer demographics and sales, making these two topics the main focus of the dashboard in Power BI. The two models were added in Power BI and we matched our initial visuals with the Machine Learning models to come to four pages of interesting information for the Marketing & Sales department. When conducting predictive analyses and creating Machine Learning models, it is essential to make the analyses and the Machine Learning technology understandable for employees, otherwise all kinds of challenges arise, such as skepticism towards autonomous decision-making and privacy. This made us look at our data through a different lens, so that skepticism would be reduced as much as possible. We made sure that our created dashboards were not only understandable for ourselves and for our Microsoft Learning environment, but would also be understandable for employees of FitCapacity who do not have much experience within data technology.
Results and outcomes
From the results created by the various data visuals, a few feasible recommendations can be made to the Marketing & Sales department of FitCapacity.
Customers
From the predictive demographics of customers on the total purchases YearToDate, it seems that as of right now, the most important customers for FitCapacity are customers who have a flag at home, do not have any children, are married, own two cars and earn €25.001-€75.000 a year. These customers are young and married people who have a good job, but are not yet completely settled with children. These are the customers that bring in the most money for the company and can be considered as FitCapacity’s target group. The customers who barely buy anything from the company are the customers who only have a high school education or a graduate degree, five or three children, four or three cars, a manual or clerical occupation and yearly income greater than €75.000. In our eyes these customers can be divided into two groups: a low educated group with a manual job and many children and a ‘high class’ group who earn a lot of money. It can be concluded that FitCapacity does not speak to these people and with new marketing efforts FitCapacity can try to include them into the company. However, the company can make a choice in this matter: focus on their target group and ensure that these customers are as connected as they can be with the company or try to be attractive to as many people as possible. We would advise the company, considering the two rising competitors, to focus on their target group and ensure that they do not lose any customers to their competitors. See figure 2 for the PowerBI dashboard where we based our outcomes on.
Figure 2. Customer information
Special offer and sales reason
With an abnormally high amount, no discount on a product is the most used ‘special offer’, which also makes it a high predictor for Total Due. An opportunity can lie here for the company, as price is the main reason why customers buy a product of FitCapacity in all territories. We would advise the company to have more discounts, to make the price even more attractive to the customers. As price is the main reason, a lot of extra sales could be attracted by the company by offering even a few more discounts for the most bought products, such as within Tires and Tubes. The company could also offer discounts to customers who in the past have bought various products from the company. This keeps the current target group close to the company and keeps them more attracted to stay with FitCapacity instead of going to a competitor. See figure 3 for the PowerBI dashboard where we based our outcomes on.
Figure 3. Additional customer insights
Territory
From the data gathered related to territory, it shows that even though most customers do not live in the US, they do make up for the highest total orders. Together with Australia, Southwest and Northwest they are contributing to the prediction of Total Due of FitCapacity. However, the Total Quantity in the US has decreased over the last few years. Interestingly, most customers live in Australia and this is also the territory with the highest predictive contribution, but they have the least amount of Total Quantity over the years. However, this Total Quantity is slightly increasing. Based on this data, we advise FitCapacity to focus their marketing initially on the US as the sales have dropped there over the years, but the customers who do buy from here have a substantial influence on the amount of sales. Thereafter, the attention should be focused on the customers of Australia to bring them closer to the company as they have potential to be of importance in the future. See figure 4 and 5 for the PowerBI dashboard where we based our outcomes on.
Figure 5. Additional sales information
Figure 4. Sales
Lessons learned
Throughout the project, we encountered various insights, challenges, and takeaways. Reflecting on our experience, we recognized the importance of understanding our data beyond surface-level. Rather than only "seeing" the data, it’s another thing to truly comprehend it. The ethical considerations that were made while analyzing the Microsoft dataset, are essential reflections and contemplations for not only FitCapacity, but can also be extended to corporate and societal environments. It came to light that understanding your data instead of only seeing it and mindlessly conducting analyses is of great importance and it is essential to consider how transparent you strive to be about the used data in your results. Moreover, understanding the structure and nature of the community you are creating models for results in responsible usage of Machine Learning and making your results understandable for your employees creates a reliable workspace with reduced chances of privacy and fairness issues. All things considered, these ethical insights enable a safe and fair environment for working with Machine Learning models for research, corporate environments, employees and society in general. This understanding allowed us to address biases, establish correct relationships, and ensure the accuracy and reliability of our results.
Collaboration and teamwork
We shared the work by having multiple meetings together in which we worked together on the project. Both of us worked on all the aspects of the project, so making visuals in Power BI and writing the report and blog. We also worked separately at home but ensured that these hours were equivalent to each other.
Future development
As we conclude our analysis and provide recommendations to FitCapacity's Marketing & Sales department, we recognize that there are several areas to explore for future research. While we identified the current target group based on the analysis of customer demographics, there is potential for further segmentation to better understand customer preferences and behavior. By analyzing additional variables such as customer interactions or preferences, FitCapacity can create more targeted marketing campaigns and personalized experiences. Although we focused on FitCapacity's market position, conducting an analysis of competitors can provide valuable insights. By comparing market share, pricing strategies, product offerings, and customer satisfaction levels of competitors, FitCapacity can identify areas for improvement and develop strategies to keep a good market position. Continuing to prioritize ethical considerations and data governance is crucial. FitCapacity can establish clear guidelines and protocols for data collection, storage, and usage to ensure compliance with privacy regulations and maintain data security. Regular audits and reviews of data practices can help identify any potential biases or unfair practices.
Conclusion
In conclusion, our analysis of the dataset provided by Microsoft has enabled us to offer feasible recommendations for FitCapacity, with a specific focus on their Marketing & Sales department. By combining descriptive and predictive analysis techniques, we have gained valuable insights into FitCapacity's market position and identified areas of opportunity for the company.
Throughout the project, we emphasized the importance of ethical considerations in data analysis. We carefully selected and prepared the data. By using Power BI, we were able to visualize and understand the data before conducting predictive analyses, ensuring a comprehensive understanding of the environment in which we were building Machine Learning models.
The results of our analysis have led to several recommendations for FitCapacity's Marketing & Sales department. Firstly, we identified the target group of customers who bring in the most revenue for the company. By understanding their demographics and preferences, FitCapacity can target their marketing efforts to this group and ensure they remain engaged with the company. Additionally, we mentioned the importance of offering discounts as a sales strategy, as price was found to be the main reason for customers to choose FitCapacity's products. By implementing discounts, especially for popular products, FitCapacity can attract more customers and keep their existing ones.
Furthermore, our analysis of territory data revealed insights into customer behavior and purchasing. Despite most customers not coming from the US, they do contribute to FitCapacity's total orders mostly. Therefore, we recommend focusing initial marketing efforts on the US market, which has experienced a decline in sales in recent years. Additionally, attention should be given to customers in Australia, as they show potential for future importance.
Throughout the project, we gained insights into the responsible usage of Machine Learning models and the importance of making the results understandable for employees. By considering ethical implications, ensuring transparency, and addressing biases, FitCapacity can create a reliable and fair workspace while making use of data technology.
If this post made you interested, we encourage you to check out the following resources, as they helped us throughout our project.
Explore what Power BI can do for you
Create and use Analytics reports with Power BI
Use Automated Machine Learning in Azure Machine Learning
Feel free to reach out to sylvie.toonen@ru.nl or marjolein.kok@ru.nl if you have any questions.
Updated Jun 19, 2023
Version 1.0