Credit Karma: Quantify marketing lift using deposit trends and client acquisition metrics. Python, Pandas, Matplotlib.
Credit Karma is a personal finance platform that helps people understand and manage their credit and debt. The customers in this exercise are people who have a substantial amount of debt and because of an unexpected hardship, are no longer able to make their minimum monthly payments. Upon enrolling in the debt relief program, customers cease making payments to their creditors and instead make deposits they can afford into a new dedicated bank account. The program uses these funds to negotiate with each of the client’s creditors to settle the debt, typically at only a fraction of what was originally owed. Once a settlement has been agreed upon for an individual account, the program facilitates payments from the dedicated bank account to the creditor based on the terms of the settlement agreement. It then collects fees from the client for the individual account that was settled. Fees are earned for each account where a settlement agreement is successfully negotiated. The number of settlement agreements that can be negotiated is proportional to the monthly deposited amount. Assume industry statistics for any analysis input that you would want to use that is not provided here, including but not limited to the fee percentage collected along with ongoing cost per client between initial enrollment and program graduation.
Imagine that Credit Karma ran a recent marketing campaign to promote the value proposition of how the debt relief program helps people achieve financial freedom. Assume the cost of this campaign was $5 million. There are five months of data in the datasets provided. Let’s say campaign took place over the course of the third month. You now want to show the marketing, sales and operations teams just how successful this campaign was.
Using the three datasets given:
Attached you will find three files with the data you will need to complete the analysis.
client_data.csv: You will find data specific to fictional clients
client_id: Randomly generated unique surrogate identifier for a clientclient_geographical_region: Client geographical location in relation to U.S. Census definitionsclient_residence_status: Client residence status in relation to whether they rent or ownclient_age: Client age in relation to date of birthdeposit_data.csv: You will find data specific to the client deposit behavior
client_id: Randomly generated unique surrogate identifier for a clientdeposit_type: Delineates whether a client deposit is the scheduled record or actual recorddeposit_amount: Client deposit amount to the dedicated bank account with the programdeposit_cadence: Timing and pattern of client deposit activitydeposit_date: Deposit date for deposit typecalendar_data.csv: This is a calendar reference table
gregorian_date: This date aligns with the Gregorian calendarmonth_name: These are the designated months in the case study
Note: These datasets were created for this analytical exercise and are only intended to be used to assess the critical thinking and technical abilities of interview candidates. This data purposely does not reflect actual client or deposit information. No inferences should be made from this information in regard to Credit Karma’s client base, deposit activity, company size or company growth trajectory.
You are free to use any tool or means you would like to highlight your abilities, but we would like to see something more advanced than Excel pivot tables, for example. Not only your ability to provide the correct results is important, but also how you decide to visualize and explain your results.
Assume the following:
See the full interview round by round, with questions for each stage.