Automated Substitute Product Identification via SKU Attribute Matrix
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying substitute products are inefficient and not scalable, particularly when dealing with large numbers of stock keeping units (SKUs), as they rely on manual identification or cross-price elasticity, which is impractical for hundreds of thousands of SKUs.
Innovation Solution
A system and method that create a matrix of SKUs with associated product attributes, allowing for the identification of substitute pairs through a subset of possible product pairs, using analytical sales data to generate a model that automatically identifies additional substitute pairs, and refines the model based on user-defined relationships.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual identification of substitute products is used, then accuracy can be maintained, but scalability is lost when dealing with hundreds of thousands of SKUs
Solution Approach 1:
The patent replaces manual expert identification (mechanical human operation) with an automated computational system that uses machine learning models and algorithms to identify substitute products, enabling scalability to hundreds of thousands of SKUs while maintaining or improving accuracy
Solution Approach 2:
The system enables automatic self-identification of substitute products through data-driven models that analyze product attributes and sales data without requiring continuous manual intervention, allowing the system to scale autonomously
2Productivity
If cross-price elasticity is used to identify substitutes, then a systematic approach is provided, but it becomes impractical for large numbers of SKUs
Solution Approach 1:
The patent segments the analysis by first identifying candidate substitute pairs using product attribute similarity (grouping products by category, brand, features), then applying cross-price elasticity analysis only to these reduced candidate sets, making the computational process manageable for large SKU volumes
Solution Approach 2:
The system performs preliminary filtering of product pairs based on attribute similarity before conducting the computationally intensive cross-price elasticity analysis, reducing the search space and making large-scale analysis practical
3Reliability
If stockout data is used to identify substitutes, then real-world substitution behavior is captured, but such data is not available when supply chain avoids stockouts
Solution Approach 1:
The patent uses product attribute data and sales data as intermediary proxies to infer substitution relationships when direct stockout data is unavailable, allowing the system to identify substitutes through alternative data sources that are always available
Solution Approach 2:
The system performs preliminary analysis using always-available data (product attributes, sales data) to identify candidate substitute pairs, which can then be validated or refined when stockout data becomes available
4Measurement precision
If all possible product pairs are reviewed to identify substitutes, then complete accuracy is achieved, but the number of pairs to review becomes impractical
Solution Approach 1:
The patent segments the complete set of product pairs into manageable subsets based on product categories, attributes, and similarity metrics, allowing systematic review of only relevant pairs while maintaining comprehensive coverage of actual substitutes
Solution Approach 2:
The system uses a multi-pass approach where initial automated filtering identifies high-probability substitute pairs, then targeted review of these partial sets achieves sufficient accuracy without requiring complete review of all possible pairs
Data Source
AI summary
This disclosure includes various methods and systems for automatically identifying product substitutes based on correlating product attributes to attributes of products in a subset of possible product pairs that are identified as substitute pairs.