Digital Promotion Server Using ML for Personalized Loyalty Indicators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital promotion systems lack the ability to effectively tailor promotions based on users' diverse purchasing habits across different categories, such as automotive and grocery purchases, using machine learning to generate personalized loyalty indicators and targeted digital promotions.
Innovation Solution
A digital promotion system that utilizes a server to repeatedly obtain low-frequency and high-frequency purchase data from user devices, employing machine learning to generate a current loyalty indicator and create tailored digital promotions, which are then communicated to the user device, with the system capable of determining communication preferences and promotion values based on purchase patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional digital promotion systems are used, then promotions can be distributed to users, but the promotions cannot be effectively tailored to users' diverse purchasing habits across different categories
Solution Approach 1:
The system segments purchase data into different frequency categories (low-frequency and high-frequency purchases) and different product categories (automotive, grocery, medical, etc.). This segmentation enables targeted analysis of specific purchasing behaviors without requiring the system to process all purchase data uniformly, thereby improving personalization capability while managing system complexity.
Solution Approach 2:
The system dynamically adjusts the loyalty indicator calculation by incorporating both low-frequency and high-frequency purchase data with different weights and time considerations. The machine learning model continuously adapts to changing purchasing patterns, allowing the promotion personalization to evolve over time without requiring complete system redesign.
2Measurement precision
If machine learning is used to generate personalized loyalty indicators, then promotion relevance to users is improved, but data processing requirements and computational resources increase
Solution Approach 1:
The system extracts only the essential features from purchase data that are most relevant to loyalty assessment, such as purchase frequency, recency, and category preferences. By extracting and focusing on these key features rather than processing all raw data, the system achieves accurate loyalty indicators while reducing computational power requirements.
Solution Approach 2:
The system processes purchase data at different frequencies - low-frequency purchases (automotive, medical) are analyzed less frequently, while high-frequency purchases (grocery) are analyzed more frequently. This partial processing approach maintains measurement precision for loyalty indicators while optimizing computational resource allocation based on the actual impact of different purchase types.
3Reliability
If the system processes both low-frequency and high-frequency purchase data, then user loyalty measurement becomes more comprehensive, but data collection and processing time increase
Solution Approach 1:
The system implements periodic data collection and processing schedules that differ based on purchase frequency. Low-frequency purchases are processed at longer intervals, while high-frequency purchases are processed more frequently. This periodic action ensures comprehensive loyalty measurement across different purchase types while minimizing unnecessary processing time for low-frequency categories.
Solution Approach 2:
The system pre-processes and categorizes purchase data as it is collected, organizing it into low-frequency and high-frequency buckets with appropriate metadata. This preliminary action prepares the data for efficient batch processing later, reducing the time required for comprehensive loyalty analysis while maintaining measurement reliability.
Data Source
AI summary
A digital promotion system may include a user device and a digital promotion processing server. The digital promotion processing server is configured to repeatedly obtain low-frequency purchase data associated with a given user for a first category of purchases, and repeatedly obtain high-frequency purchase data associated with the given user for a second category of purchases different than the first category of purchases. The high-frequency purchase data may represent a greater number of purchases made in a given time period relative to the low-frequency purchase data. The digital promotion processing server is also configured to use machine learning to generate a current loyalty indicator based upon the repeatedly-obtained low-frequency purchase data and based upon the repeatedly-obtained high-frequency purchase data, and generate a digital promotion based upon the current loyalty indicator and communicate the digital promotion to the user device.