Imagine your team is responsible for churn management at a media company with a subscription-based business model. The company offers an annual membership that gives members the right to use its services (both online and offline). One month before a customer's membership expires, the company sends out a renewal letter each year. If the membership is not renewed, the benefits can no longer be received.

In the next three months (May, June, and July), 2,306 customers are up for renewal. The company is contemplating sending customers a "thank you" gift with their renewal letter. However, there are no sufficient funds to send a gift to all customers. There are enough funds to send 930 gifts, and it is your job to select the customers that will receive gifts.

The company has made available a dataset that includes customers who were up for renewal in the past three months (February, March, and April). For each customer, you know:

-for how many years the customer has been a member (tenure),

-whether the customer attended an event in the past year (attendance),

-whether the customer downloaded any online content in the past year (download),

You also know whether the customers that were up for renewal in the past three months decided to renew or not.

The company's only goal is to maximize the likelihood that a membership will be restored. They only care about differences in renewal rates. You can ignore potential adverse spillover effects (e.g., customers talking to each other about whether they received a gift or not), and you can skip the cost of the gift for now.

How should the company select customers to target? Which customers should receive a gift?

https://docs.google.com/spreadsheets/d/1o58OonzAvS0fFj_jQXu_1_gSX9530j0B-Z_lJX6-MKk/edit#gid=319159061