Abstract:

Over the past two decades, digital platforms have become integral to meeting the diverse needs of consumers. Given the wide choice of platforms, organizations employ various mechanisms to retain them. These include person-alized recommendations, alerts about offers and rewards, reminders to rate items, subscribing for premium services, etc. However, excessive notifications could overwhelm users, resulting in dissatisfaction and loss of user base. To address this problem, we introduce NudgeX, a framework for intelligent nudges to bal-ance user engagement and user experience.

NudgeX, which is in its beta stage, orchestrates and optimizes the timing and type of nudges across three key user touchpoints: onboarding, performing inter-actions, and feedback provision. The nudges are initiated based on Triggers, de-fined by user actions, predefined scenarios, or periods of inactivity. The users are segmented as Blue (new or recently signed up users), Red (inactive or less en-gaged users), or Green (active users) based on their level of engagement. NudgeX is novel in its approach to integrate various machine learning models powered by Decision Tree classifier, Singular Value Decomposition, Random Forest Classi-fier, Linear SVC, and Cosine Similarity to provide personalized nudges, display recommendations, and predict user item ratings. The framework also provides explainability to provide the rationale behind the selection and timing of the nudges delivered.

The models are evaluated based on accuracy and F1 score. In a simulated envi-ronment with 300 users and 2,000 transactions, NudgeX performed with accuracy ranging from 0.92 to 0.99 across scenarios.

Keywords:

Digital Nudges, Explainability, Reuse Repositories, User Experi-ence, Recommender Systems

Conference Published in: 7^th International Joint Conference on Advances in Computational Intelligence IJCACI 2023