Task 1. Which major elements best describe the value of a XING user?
Start by explicitly defining up to five major elements of how users (either mobile or non-mobile) contribute value to the platform. Then, give these elements proper and catchy labels (e.g., “monetary value”).
An excellent start is the paper Anderl, März, and Schumann (2016): Nonmonetary customer value contributions in free e-services (particularly pages 5-9; also provided on Moodle with the kind permission of the authors) on the value of non-paying users. However, Anderl et Al. (2016) framework is a rather general one. Therefore, you may also reflect on the specific XING and New Work context, as presented by Michael Horn.
List a catchy name for up to five major elements and briefly describe them. What is specific about the XING case in comparison to Anderl et Al.’s framework (2016) on page 5-9?
Task 2. Without looking at the data, explain your expectations of whether mobile users will be more or less valuable compared to non-mobile users for each major element
This step is about forming and explicitly stating your expectations on how mobile users differ from non-mobile users for each of the (up to) five previously identified major elements. Then, make an argument why you believe that mobile users are either (1) more valuable, (2) equally valuable, or (3) less valuable for each of the identified major elements from step 2. This step does not require any data analysis – it is about explicitly articulating your intuition.
Task 3. Go through the data and select up to two variables that best represent each of the previously identified major elements
You may find some additional information on the provided data in the last slides of Michael Horn’s slide deck (PDF). For this step, no data analysis is required. Just inspect (= eyeball) the data to see whether your proposed columns can represent one of the proposed major elements from step 2. You may also assess whether the value provided in the data makes sense to you and whether the column does not have too many missing values. At a maximum, you could select ten columns from the data. These columns may already exist in the data, or you formed them by combining multiple columns.
So, which variables have you chosen to represent the major elements? List them and provide a short reasoning why.
Task 4. Now test whether the business data supports your intuition
Run the following analysis for each of the variables in step 3 (up to ten). There are two ways to go, and you can decide on just one of them that works best for you (we added a digital learning unit for more information on the difference: Link):
Run multiple student t-tests in PSPP
In each of the tests, use one variable from step 4 as the dependent variable (Y) and the column “MOBILE_USER_DUMMY” from step 1 as the categorizing independent variable (X).
Use the regression module in Excel
Because running a student t-test is not convenient in Excel, you may approximate the finding above by running multiple regressions instead of multiple student t-tests. In each regression, use one variable from step 3 as the dependent variable (Y) and the column “MOBILE_USER_DUMMY” from step 1 as the categorizing, independent variable (X) and a constant. The videos on the deep dive on regression may help you here. For simplicity, you may even use linear regression (and not, e.g., logistic regression) to test the significance if the dependent variable is binary.
Reflect on the direction of the effect and the level of significance.
Summarize the results from step 4 (see “XING Premium Challenge_OMBA – Instructions”) in one meaningful table and interpret the findings. Pay attention to the signs of the parameters and their level of significance. Are all your expectations met, and if not, what could be the reason?
Task 5. What drives a user’s social capital?
The previous analysis just used one independent variable (MOBILE_USER_DUMMY), and you tested its role on various dependent variables that represent the value of a customer. We now concentrate on users’ social capital, often defined in the context of XING as the professional relationships, connections, and networks the user builds and maintains on the platform. This includes trust, shared norms, and mutual support within their professional network, which can lead to career opportunities, collaborations, and knowledge sharing. Despite the definition’s complexity, scientific literature often operationalizes social capital by measuring a user’s number of contacts (in our case, listed in column “num_contacts”). For more information on social capital, see our latest publication in Management Information Systems Quarterly (optional, provided on Moodle).
Your task is to examine what drives social capital.
Plot a histogram of the “num_contacts” variable (this is possible in Excel). Does this histogram look like a normal distribution?
Create a new column “ln(num_contacts+1)”, where you take the natural log of (num_contacts +1). The +1 is important to obtain a valid number, in case the information in “num_contacts” is zero. Now, also plot this histogram. Decide which of the two variables (“num_contacts” or “ln(num_contacts+1)”) better approximates a normal distribution.
Let’s create a final variable called “social_capital”, which is either “num_contacts” or
“ln(num_contacts+1)”, depending on whether it better approximates a normal distribution.
Use “social_capital” as the dependent variable for the following step: Select up to ten meaningful additional columns as independent variables that potentially explain why some users have higher social capital than others. Please include MOBILE_USER_DUMMY as one of the ten variables.
Run a regression and interpret your insights.
Plot the two graphs and list your regression results. What do you learn?
Task 6. What is the link between social capital and the probability of subscribing to Premium?
For the final analytical task, concentrate on the likelihood of subscribing to a paid premium membership depending on the user’s social capital. Again, this task asks for your intuition first and then challenges you in running the respective analysis and interpreting the results.
The more contacts a given user has, the higher his/her social capital in the network. Without looking at the data, provide arguments on whether users with low social capital (e.g., the 10% of users with the lowest number of contacts) are likely to upgrade to Premium compared to those with high social capital (e.g., the 10% users with the highest number of contacts) or those ranged in the middle. Explain your reasoning.
Investigate the relationship between Social Capital and the decision to buy Premium.
Calculate the deciles of the new variable (i.e., 10% intervals). See, e.g., https://www.statology.org/deciles–in–excel/. Then, create another new variable, “Decile”, that places each data value into a decile (from 0 to 0.9; see the link above).
Within each decile, determine the percentage of users who bought Premium (this is equivalent to the observed purchase probability within each decile).
Create a graph that plots the decile membership on the x-axis and the percentage of users who bought Premium on the y-axis.
Plot the graph and list the probability of subscribing to premium in each decile in a table. Is this observation in line with what you anticipated in 6a.? What could be the reason?
Task 7. Derive actionable recommendations
Reflect on the analysis in the previous steps and provide actionable recommendations to the management of customer relationship management.
Anti-plagiarism rules
You are expected to work on the case and do your calculations all on your own.
Cases and case solutions may not be shared with students from your class, future classes, or students from other WHU programs. If you circulated solution blueprints, the case and the case method would be effectively ruined for learning at WHU and beyond. Respect for copyright and intellectual property is among the highest academic imperatives.
By signing up for the WHU program, you have authorized WHU to scrutinize your submissions and case write-ups for plagiarism. We use plagiarism check tools to scan for overlaps with submissions of other WHU students. Using material from other students will lead to a severe reduction of points or, in the worst case, be treated as cheating.
Using material from other students and sending material to other students will be considered a violation of this code and deception in the sense of WHU’s examination order. The reduction of points or, in the worst case, the treatment of cheating also applies to the person who provided the information. By giving material to other students, particularly as one of the better students, you might personally feel like a compassionate helper. You might even get the recipient’s thanks and social praise for your “altruism” and “generosity”. But what you are doing is undermining the system’s integrity and hampering the recipient’s professional maturation.