Predictive Sentiment Analysis and Proactive Service Intelligence: Implications for Long-Term Customer Loyalty and Retention-A Scoping Review
Rahmat Kofoworola Sanusi *
University of West Georgia, Carrollton, United States.
Temiloluwa Paul Adetola
The NorthBridge Insight, Linnamäe 45, Tallinn, 13911, Estonia.
Abimbola Serifat Oreoluwa
Department of Cybersecurity, Nexford University, Washington, DC, United States.
Sampson Boateng
The Roux Institute - Northeastern University, Village Health Works, United States.
*Author to whom correspondence should be addressed.
Abstract
Background: Predictive sentiment analysis and proactive service intelligence are increasingly used to identify deteriorating customer relationships and support targeted service responses, yet evidence is distributed across marketing, customer relationship management, service research and business analytics.
Objective: This scoping review maps the evidence linking sentiment analysis, predictive customer analytics and proactive service with customer loyalty and retention and develops an integrated conceptual synthesis.
Methods: The review was guided by established scoping-review methodology and PRISMA-ScR. Nine multidisciplinary databases were searched, supplemented by backward citation searching and citation tracking. Evidence was charted by study design, industry, technology, data type, analytical method, customer-intelligence construct, service mechanism and relationship outcome.
Results: A total of 515 records were identified, 60 full-text reports were assessed and 32 sources were included. The mapped evidence indicates that textual sentiment signals can add predictive value when combined with behavioural and transactional data; AI-enabled customer analytics can support churn-risk identification and personalised service; and proactive service may translate customer intelligence into timely interventions. However, prediction accuracy should not be equated with intervention effectiveness, and direct longitudinal evidence for durable loyalty and retention remains limited. Trust, privacy, explainability, human oversight and responsible data use emerged as important boundary conditions.
Conclusions: The literature supports a conceptual pathway from customer data to sentiment intelligence, predictive risk assessment, proactive intervention and relationship outcomes, but the pathway should be interpreted as an evidence-informed synthesis rather than a validated causal model. Future research should prioritise longitudinal and experimental evaluations of intervention effects and heterogeneous treatment response.
Keywords: Predictive sentiment analysis, proactive service intelligence, artificial intelligence, predictive analytics, customer loyalty, customer retention, customer churn, customer experience, machine learning, customer relationship management