Subscription churn

Nayimoni

Hello, I have a model that has gym subscriptions of 1-3-6-12 months that churn and get renewed same month of expiry or later (1,2, n months later) with each row representing a member ID and its details. A member that renews is entered as new row for same ID. I am trying to figure out how to use survival platform to:

 

- determine renewal rates (proportion of those who renew one time, two times etc.) for each subscription type

- determine customer lifetime value by determining average tenure (taking into account lag to renewal)

 

Any help how to use JMP Pro please to get this.

 

Thanks

dale_lehman

It would help to see a sample of the data, but from your description I'd try to approach it this way:  the response variable would be the time to nonrenewal and there would be a censoring variable indicating whether a person has churned or not (those that renew are censored, since they might fail to renew later on).  One of the independent variables of interest would be the length of subscription.  This would be a fairly standard survival analysis except for one complicating factor - your observations apparently show repeated renewals.  In other words, you have sufficient time to have observed some people not renewing subscriptions, some renewing, and some renewing more than one time (and perhaps eventually not renewing).  Modeling the repeat renewal behavior seems more complicated to me, but I would start with the simple model that has each observation either churning or not, and the response variable being the time to nonrenewal (or the total time subscribing for those that only show renewals).  The factors of subscription lenghts and perhaps number of subscription decisions then might provide an initial model to look at.