How can I predict changes in customer lifetime value?

This is of course the crux of the ROC idea. To calculate the long-term customer value created by its actions, a firm must be capable of somehow predicting today that a customer lifetime value will increase or decrease. We¹ve identified four types of ³leading indicators² of changes in lifetime value-variables that can be measured today and indicate a likely change in the future behavior of a customer or customer group. Lifestyle changes include events like marriage, pregnancy, retirement, or divorce, which can obviously change a customer¹s future behavior. Then there are the drivers of a company¹s LTV equation itself‹customer-churn rate, repeat purchase frequency, and so forth. Third are behavioral cues. Customers who change their purchase patterns, frequency of contact, or their payment frequency are giving you a clue to their future behavior. The fourth leading indicator is attitude.
Attitudes precede behavior, so changes in customer satisfaction levels, willingness to recommend, or other such attitudes are all signals that communicate likely changes in the future behavior of customers.