Churn management is a top priority for most businesses as it directly ties to firm profitability and value. Companies in a variety of sectors (e.g., telecommunications, insurance, healthcare, online subscriptions) increasingly start managing churn proactively, generally by detecting customers at the highest risk of churning and targeting retention efforts towards them. While a significant amount of work has been dedicated to developing churn prediction models that accurately assess customers' risk of churning, no work has investigated whether it is indeed optimal to target individuals who are at the highest risk of churning.
In this paper we challenge the most common practice for proactive churn management and claim that, when the main goal is to select customers for preventive retention efforts, identifying customers at high risk of churning does not suffice to drive the firm's decisions. We argue that because customers respond differently to the firm's intervention, a firm should not target those with the highest risk of churning but those with the highest sensitivity to the intervention. Consequently, we propose a new approach for proactive churn management that leverages the firm's capabilities by running a retention pilot, identifies the observed heterogeneity in the response to the intervention, and selects target customers based on their sensitivity to the intervention, hence ensuring that the retention efforts are not futile.
Combining data from two field experiments with machine learning techniques, we empirically demonstrate that the proposed approach is significantly more effective than the standard practice of targeting customers at the highest risk of churning. We show that the same retention campaign would result in a further reduction of up to seven percentage points in churn rate if the focal firm had followed our proposed approach (instead of the industry standard). Contrary to common wisdom, we consistently find that customers identified as being at the highest risk of churning are not necessarily the best targets for proactive churn programs. These findings suggest that retention programs can be futile not because they offer the wrong incentives, but because firms do not apply the right targeting rule.