Push Notification Quality Scoring Using App Usage History
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the mobile app environment, quality scores based on predicted click-through rates are less effective for selecting push notifications, as users cannot navigate to other apps without installing them, and existing methods do not adequately consider usage patterns across multiple apps.
Innovation Solution
Collecting and analyzing mobile device application usage history to adjust quality scores for push notifications, using a machine learning algorithm to determine the most relevant notifications to send based on user behavior, and auctioning opportunities among candidate notifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If quality scores are based on predicted click-through rates, then ad distribution effectiveness is improved, but relevance to user context deteriorates because existing methods do not adequately consider usage patterns across multiple apps
Solution Approach 1:
The patent changes the parameters used for quality scoring from generic predicted click-through rates to context-aware scores that incorporate mobile application usage history, trigger events, and sequential usage patterns. This allows the system to maintain ad distribution effectiveness while significantly improving ad relevance to user context by considering what apps the user has recently used and the sequence of their usage.
Solution Approach 2:
The patent implements feedback mechanisms by collecting and analyzing mobile application usage history and trigger events to continuously refine quality score calculations. The system uses observed usage patterns and sequential app usage data to adjust and improve the relevance of push notifications over time, creating a feedback loop that enhances both reliability and measurement precision.
2Ease of operation
If push notifications are sent without considering application usage history, then notification delivery is simplified, but user engagement deteriorates due to lower relevance
Solution Approach 1:
The patent applies preliminary action by collecting and analyzing mobile application usage history and identifying trigger events before sending push notifications. The system pre-processes usage data to understand user behavior patterns, so when a notification needs to be sent, the quality score is already optimized based on this preliminary analysis, maintaining simplicity while significantly improving user engagement.
3Measurement precision
If usage history is collected across multiple applications, then notification relevance is improved, but data processing complexity increases
Solution Approach 1:
The patent extracts only the most relevant features from mobile application usage history, such as sequential usage patterns and trigger events, rather than processing all raw usage data. This extraction approach maintains high notification relevance by focusing on key behavioral indicators while reducing data processing complexity by eliminating unnecessary data elements.
Data Source
AI summary
Mobile device application usage history is collected across a plurality of mobile device applications. A mobile device application usage trigger event is received. For each of a plurality of candidate push notifications, a quality score adjustment is determined as a function of the collected mobile device application usage history and the trigger event. An opportunity to push a notification to the first mobile device is auctioned based on the adjusted quality scores. The notification of the auction winner is pushed to the first mobile device.