I would like to get a second opinion of the problem of assigning engagement scores to online news readers.
Currently, I build a churn model as required by the company, that mainly predicts the likelihood of an online news reader to stop its monthly subscription. The model has variables like: Time as customer, number of read articles in past week, total read time in past week, type of product, type of subscription, device types, etc.
Now, I was asked to build an engagement index, to assign for each online news reader, an engagement score, potentially real time prediction if possible, so that the editorial can understand the engaged audience.
I feel that modeling engagement score model, is the opposite of measuring churn risk. If churn is low, then user is engaged.
Do you think there is a need for an engagement score model if I already have a churn model? If so, how to best handle this second problem of news reader engagement?