Businesses need to understand not just what a customer has been worth to date, but what a customer will be worth in the future, whether that is in the next quarter, next year or next five years or if that is truly over the lifetime of the customer. First, this allows businesses to prioritize their most valuable customers. Second, by comparing expected lifetime value to the value provided to date, businesses can work to cultivate those customers with significant future value.
Predicting customer future value allows for the prioritization and segmentation of customers.
Customer lifetime value
Businesses need to understand what exactly they want to know. True lifetime value is tough to model (most customers have not completed their lifetime, leading to a censoring problem) and further treating revenue that may not arrive for decades as equivalent to revenue in the next year may not be appropriate (as it ignores the time value of money).
To predict future value, it is necessary to pick a duration to capture the value and calculate the total value the customer provided in that period (this may require merging multiple datasets if the company has multiple products with CRM systems dedicated to each one). This data needs to be merged with both the customers demographic information and account age from the CRM system as well and their behavior patterns (purchases, marketing interactions, support interactions) from the period prior to the one used to calculate the value. All this data needs to be cleaned, joined and transformed into valuable ML features before going into model training. This pre-modeling prep process can be frustrating and time consuming. We are here to help.
Build machine learning models to predict the target as a function of the independent variables. Use model interpretability packages to evaluate the impact of the independent variables on the prediction.