![“Walking a Fine Line: Customer Retention in Mobile App Targeting” by Ms. Xingying Hao](https://www.hkubs.hku.hk/wp-content/uploads/fly-images/2430/marketing-847x560-c.png)
“Walking a Fine Line: Customer Retention in Mobile App Targeting” by Ms. Xingying Hao
Ms. Xingying Hao
Ph.D. Candidate in Marketing
McCombs School of Business
The University of Texas at Austin
We address a common pitfall largely neglected in mobile app management: over-targeting. Theoretically, we conceptualize three different types of over-targeting commonly seen in mobile push marketing: (1) over-precision (2) over-exposure and (3) over-saturation. We utilize mobile tracking technologies to empirically investigate how the targeted push notifications, based on a user’s past behavior (i.e., behavior-based push) or based on a user’s current location (i.e., location-based push), can affect the mobile user’s churn, retention, re-engagement, and privacy choices in the context of mobile app CRM (customer relationship management). We propose a Bayesian hidden Markov model, and we apply it to a unique mobile dataset containing granular user-level data. In addition, we model the nonlinear effect of push notifications to help mobile app publishers answer a question for designing the effective push scheduling system: At what point does a mobile app reach the over-targeting threshold for each mobile customer? Our results show that push notification scheduling can be a balancing act between retention and disengagement. Too few pushes may lead to users abandoning the app and its services, while too many pushes may result in an overload of information and privacy concerns. Most importantly, these thresholds heavily depend on the current hidden customer relationship state (i.e., should be targeted, privacy concerned, and intrusiveness sensitive) as well as the usage level.For instance, considering the frequency, two to four behavior-based push notifications per week are usually the most effective. But, if it goes up to seven pushes per week, most customers are repulsed. We further demonstrate that over-targeting has spillover effects, in which excessive behavioral targeting may trigger privacy concerns, thus compelling users to disable location services. In addition, our short-term and long-term effects offer behavioral explanations for the three types of over-targeting that we describe. In a managerial context, our simulation of push strategies has implications for designing well-timed and well-targeted push notifications that maximize customer retention and re-engagement. For instance, we find that sending one more (one less) notification than the frequency threshold per week may cause a firm to lose up to 6.2% (4.1%) of its customers and 11.5% (7.1%) of its active customers after five weeks.