User Engagement Prediction via Mobile Sensor Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advertisers face challenges in accurately measuring user engagement with mobile advertisements, particularly in determining the time spent by users, which affects optimization and fraud prevention, due to varying content lengths and user distractions.
Innovation Solution
A system that collects data from user sessions, including content type, user interactions, and sensor data, to predict user engagement levels by applying engagement predicting rules, allowing for personalized content presentation and fraud detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If advertisers measure performance based on number of views, then they can estimate users reached by advertisement, but they cannot accurately compare different advertisements with varying content lengths or determine user time spent engaging
Solution Approach 1:
The patent introduces an intermediary system comprising sensors (accelerometer, gyroscope, proximity sensor, light sensor) and processing logic that mediates between the advertisement display and the measurement of user engagement. This intermediary collects device state data, analyzes user behavior patterns, and computes engagement metrics, thereby enabling precise measurement of user time spent and engagement levels without directly modifying the advertisement content or display mechanism.
2Ease of operation
If mobile users use their device while engaging in other activities, then they are frequently distracted or disengaged, but advertisers are unable to determine user distraction or disengagement during advertisement presentation
Solution Approach 1:
The patent implements feedback mechanisms where sensor data (accelerometer, gyroscope, proximity sensor, light sensor) continuously monitors device state and user behavior during advertisement presentation. This feedback loop enables the system to detect changes in device orientation, movement patterns, and environmental conditions that indicate user distraction or disengagement, allowing real-time adjustment of engagement assessment without interfering with the user's multi-tasking activities.
3Productivity
If advertisers cannot accurately determine user engagement metrics, then they are limited in optimizing advertisements, but implementing comprehensive tracking increases system complexity
Solution Approach 1:
The patent leverages the universal nature of mobile device sensors (accelerometer, gyroscope, proximity sensor, light sensor) that already exist for other functions such as navigation, camera operation, and display adjustment. By repurposing these multi-functional sensors for engagement measurement, the system avoids adding dedicated tracking hardware, thereby minimizing increased device complexity while enabling comprehensive user engagement analysis for advertisement optimization.
4Reliability
If publishers inflate metrics without showing advertisements to real users, then click fraud occurs, but detecting such fraud requires accurate engagement verification
Solution Approach 1:
The patent enables the mobile device itself to self-verify engagement by using its own sensors (accelerometer, gyroscope, proximity sensor, light sensor) to collect evidence of actual user presence and interaction during advertisement display. The device autonomously generates engagement metrics based on its sensor data, eliminating reliance on external tracking mechanisms that could be manipulated by publishers. This self-service approach provides reliable fraud detection by capturing genuine user behavior data directly from the user's own device.
Data Source
AI summary
Systems, methods, and computer-readable storage media for determining user engagement levels during a presentation of content. The system first collects data associated with a user session at a client device. Next, the system predicts a user engagement level during the user session by applying an engagement predicting rule to the data. The system can predicts respective user engagement levels for various segments of the presentation by applying one or more engagement predicting rules to the data. The system then presents invitational content based on the user engagement level.