Stay-Contextual Trigger: A Confidence-Based Algorithm for Energy-Efficient and Behaviorally Grounded Mobile Notifications

Context-aware mobile applications often trigger notifications based solely on geofence entry, leading to high battery consumption and disruptive alerts during transient movement. To address this, we propose the Stay-Contextual Trigger (SCT) framework, which gates high-power GNSS activation behind low-power activity recognition and enforces a context-adaptive dwell-time requirement before delivering any notification. We evaluate SCT through a 14-day field study with 60 compliant Android users (Per-Protocol analysis after attrition filtering), comparing it against standard geofencing in four real-world outdoor zones. The SCT system computes a Contextual Readiness Score via a Confidence-Based Weighted Algorithm (CBWA), dynamically adjusting the weights of activity, proximity, dwell time, and behavioral history based on location semantics. Results show that SCT reduces daily battery drain by 40.3% (from 19.30% to 11.52%), increases notification acceptance from 32.5% to 89.4%, and achieves a System Usability Scale score of 81.25, an 18.8-point improvement over the control group’s 62.42. The high acceptance rate (89.4%) further validates sustained user engagement, ensuring metrics reflect consistent usage. These findings demonstrate that behavioral readiness, not mere proximity, is essential for sustainable and user-respectful mobile systems.