I'm doing a practice problem which goes something like this. It is homework-style but not exactly. I'm doing some practice problems myself.
Envato is a loyalty solutions which rewards their customer with points for specific purchases. The more points Envato issues, the more profits they make. Customer use these points to redeem rewards later.
Envato recently ran a campaign to provide free passes to a play to their customer (with no purchase or point redemption necessary). The campaign was sent to a test group but not to control group. How would you measure the success of the campaign?
I'm having a hard time to understand this problem from business perspective than from statistical perspective. Given the below data, what should I be testing for?
Basically, these are some important columns -
Account Balance - Current point balance of the customer
Control Flag - Whether customer is part of test or control group
Campaign Participant - If customer was in test group, did they actually go to the play i.e. participated in the campaign
Earnings During Campaign - Points earned while the campaign was taking place (roughly a month)
Post Campaign Points - Points earned after the campaign was over (roughly the same time period i.e. a month)
Email active 120 Days - Whether customer checked email offers after campaign
Online active 120 Days - Whether customer logged into the website after campaign
Customer Age, Tenure, Gender, etc
The problem doesn't state what should be the DVs or IVs or how are we really measuring the campaign success. It just asks how would you measure the success of the campaign given that the company makes profit every time points are rewarded to customers.
I felt a MANCOVA with response variable as "Earnings During Campaign" and "Post Campaign Ponts" , independent variable as "Control Flag" and "Campaign Participant" with others as covariate would be the right approach. But then it seems, I should do a separate test with "Campaign Participant" also.
