As firms recognize a Market Research Online Community or a Consumer Consulting Board as a valuable resource for integrating external consumer knowledge into research processes, they increasingly ignore temporal interaction borders and aim for long-term collaborations. However, in the pursuit of a structural, ongoing community, moderators face enormous challenges, especially due to members’ unconstructive behavior. Member disengagement, whether in the form of passive interaction and/or low-quality contributions, produces a shallow community with minimal activity and rotten community with inferior content, respectively.
Yet communities tend to be characterized by an abundance (volume) of quickly expanding (velocity) multimedia (variety) member-generated data, suggesting opportunities to leverage this big data environment to establish healthy communities. By introducing predictive analytics to communities, this study proposes a novel approach to predict and prevent member disengagement. Predictive analytics uses statistical/machine learning techniques and historical data to predict future events according to prediction models. Behavioral data (RFM) and linguistic variables (narcissism and positive emotionality) are used as disengagement drivers on the member, moderator and community level. Moreover, preventive actions for the moderator are explored to prevent member’s unconstructive behavior from negatively impacting the community. This study offers previously unseen advantages to community managers of research communities. First, member disengagement prediction enables proactive community management. Second, predicting two disengagement types separately allows community managers to take appropriate actions. Third, insights into the drivers reveal that community managers must manage the community by acknowledging the individual user and guaranteeing a positive atmosphere to prevent disengagement and build healthy ongoing research communities.