The client tasked MCG with increasing customer loyalty and reducing churn, which was eroding their profitability.
MCG developed a predictive model to identify the residential customers most likely to churn and the product offerings customers were most likely to add or drop. The extended model incorporated both external census block and advertising data. MCG’s model linked survey data (what customers say) to behavior data (what they do).
Unlike the client’s previous, ineffective systems, the model developed by MCG incorporated transaction-level information about customers such as initial product acquired, time since last transition (purchase or drop), and history of product adds and drops leading to current product portfolio.
To provide actionable data on the client’s customer loyalty, the model used causal graph modeling of survey data to show how stated loyalty intent changes as various service dimensions are improved.