Reasonable approach for modelling churn (survival) and choice of intervention campaign (multinomial regression)? I've only recently moved into customer analytics, and would love to get some advice around designing a reasonable approach to modelling my data. I want to be able to predict customer churn (that is, predict if individual customers are going to leave our service, a binary outcome) based on customer attributes (some are constant, some are due to recent account activities) but also choose the timing, campaign type and delivery method (e-letter, physical letter) of a retention missive from the company.
I feel like this needs to be broken into two models, the first being a survival model to predict when a customer would leave, and a multinomial logistic regression to identify the campaign and delivery type.
Does this sound reasonable? Any suggested reading would be appreciated.
 A: I have recently done this kind of customer analytics and generally I would say it is wise to split this problem into two parts. 
But you do not necessarily need to apply survival analysis for churning since you can think it as a panel problem where initial customer set is observed and then churned customers are observed during fixed time period. Target variable could be binary and not survival in days/weeks/months since initial panel time period.   
Second part does not need multinomial logistic regression since channel via which customer is tried to turn is just an categorical variable where one class is baseline to which other classes are compared. Same is true for campaign.   
For us telephone is baseline and other methods are compared against it. Telephone is of course also the most expensive turning method to use. 
A: A good introductory source for survival models is Paul Allison's Event History and Survival Analysis published by Sage. 
I am not too clear on exactly what you would like to do in musing the multinomial logistic model. If you are trying to choose the best retention method to chose based on their characteristics then that is one place to start. 
