Customer Health Score



Hello AV readers,

I am using some metrics to calculate customer health score(composite score), which basically measures how well the customer is using our product aka. customer adoption.However, to be able to apply some of the ML techniques I am not sure what is the best approach?

What should be my x/y variables for below cases ?

For eg-
1.) Let’s say if I want to establish threshold/target for a customer who has already in their half of their contract period i.e. 50% into their term v/s a customer who is in 20% of their term.

2.) How to do sensitivity analysis i.e. which metric moves health score needle in positive/negative direction?

3.) If overall customer score is high/low, how can I translate that to , whether a customer is going to renew/churn respectively ?

Please let me know.