Termin: 22. Februar 2018
Vortragstitel: Proactive Retention Management in Retailing: Identifying, Predicting, and Preventing Partial Defection
Referenten: Prof. Dr. Arnd Huchzermeier und Daniel Ringbeck
Zusammenfassung: Maintaining strong customer relationships is a priority for retailers because it costs less to retain an existing customer than to acquire a new one. In this paper we develop a framework for proactive retention management based on data from a German hypermarket chain. First, we propose a novel definition of partial defection, in noncontractual settings, that clearly distinguishes it from short-term volatility. Our approach enables the retailer to spend less on marketing overall and to use resources for targeting the customers most likely to defect. Second, we conduct a thorough benchmarking of prediction algorithms to forecast the likelihood of a customer to ”churn” and also evaluate methods for overcoming the challenge of high class imbalance – a typical phenomenon in churn prediction. Our best model accounts for 54% of the revenue that might be lost owing to churn. Third, we formulate a retention campaign optimization model that considers factors such as expected value loss, response probability, and incentive costs. Our analysis is based on a large data set comprising the transactions of nearly 20,000 customers over a 30-month period. We undertake a numerical study and find that, under realistic assumptions, our approach could yield a 90% return on marketing investment.
Referenz: Smirnov, Dmitry; Ringbeck, Daniel; Huchzermeier, Arnd (2018): Proactive Retention Management in Retail: Identifying, Predicting & Preventing Partial Defection. WHU – Otto Beisheim School of Management, 36 p.