Push Notification Targeting via ML Conversion Probability

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional push notification systems waste network bandwidth and system resources by sending unsolicited notifications to all users, resulting in low conversion rates as they do not account for user interest in the items promoted.

Innovation Solution

A machine learning-based system that utilizes historical transaction data to calculate conversion probability values and confidence indicators for user-item pairs, selectively transmitting customized push notifications to users likely to purchase specific items, optimizing resource usage and conversion rates.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If push notifications are sent to all users in a batch transmission, then the system can reach a large audience, but network bandwidth and system resources are wasted on users who are not interested in the items

Engineering Contradiction:

Improvenumber of notifications transmittedVSAvoidnetwork bandwidth consumption

Core Design Contradiction:

Quantity of substanceVSLoss of energy

Solution Approach 1:

The patent segments the user base into distinct groups based on their purchase history and item interest. By dividing users into segments with similar characteristics, the system can target notifications only to relevant segments, avoiding wasteful transmissions to uninterested users while maintaining comprehensive coverage of the potential audience.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by customizing notification content and selection criteria for different user segments. Each segment receives notifications tailored to their specific interests and purchase patterns, ensuring that the notification content is locally optimized for each group's preferences rather than using a uniform approach for all users.

Inventive Principle:

Principle #3Local quality

2Ease of operation

If push notifications are sent to all users regardless of interest, then transmission simplicity is maintained, but conversion rates remain very low

Engineering Contradiction:

Improvenotification transmission simplicityVSAvoidconversion rate

Core Design Contradiction:

Ease of operationVSProductivity

Solution Approach 1:

The patent performs preliminary actions by analyzing user purchase history and calculating item interest scores before generating notifications. This pre-processing step identifies which users are most likely to convert, allowing the system to maintain operational simplicity in the actual notification sending while having already optimized the recipient list and content based on predictive analytics.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback mechanisms by continuously monitoring user responses to notifications and updating item interest scores accordingly. This feedback loop allows the system to learn from actual conversion outcomes and refine future notification targeting, progressively improving conversion rates while maintaining the simplicity of automated transmission processes.

Inventive Principle:

Principle #23Feedback

3Device complexity

If conventional batch transmission is used, then system resource allocation is simplified, but resources are wasted on unnecessary notification functions

Engineering Contradiction:

Improvesystem resource allocation complexityVSAvoidsystem resource waste

Core Design Contradiction:

Device complexityVSLoss of energy

Solution Approach 1:

The patent changes key parameters such as item interest scores, user segment assignments, and notification priority levels based on analytical calculations. By dynamically adjusting these parameters, the system optimizes resource allocation to focus computational and network resources on high-value transmissions rather than uniformly processing all potential notifications, reducing waste while managing complexity through automated parameter updates.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11087237B2Machine learning techniques for transmitting push notifications

Publication Date: 2021.08.10 WALMART APOLLO LLC

AI summary

Systems and methods including one or more processing modules and one or more non-transitory storage modules storing computing instructions configured to run on the one or more processing modules and perform acts of: utilizing historical transaction information to derive metric information associated with prior transactions; generating a listing of user-item pairs, each of the user-item pairs identifying a user and an item; executing a machine learning model that is configured to generate a transmission list for sending push notifications; generating a transmission list by selecting user-item pairs based on the conversion probability values and the confidence indicators that are assigned to the user-item pairs; customizing content for the push notifications to include information for items identified by the user-item pairs included in the transmission list; and transmitting the push notifications to the users identified by the user-item pairs included in the transmission list. Other embodiments are disclosed herein.