Merchant Classification via Payment Activity Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing merchant classification system based on merchant category codes (MCC) is flawed due to self-reported inaccuracies and inability to capture contextual information, leading to misclassification and inefficient fee structures, as well as inaccurate tax reporting and customer transaction reflections.
Innovation Solution
Implementing machine learning models, such as Latent Dirichlet Allocation and Random Forest Classification, to analyze situational features from payment activity data, enabling richer merchant insights and reclassification into JTBD clusters, which can dynamically adjust based on real-time and historical data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If self-reported merchant category codes (MCC) are used for classification, then the classification process is simple and quick, but the accuracy and reliability of merchant classification deteriorates
Solution Approach 1:
The system continuously monitors payment activity data and uses it to provide feedback on merchant classification accuracy. By analyzing transaction patterns, the system identifies mismatches between self-reported MCC and actual business operations, then triggers reclassification processes to correct inaccuracies, creating a closed-loop feedback mechanism that improves classification reliability over time
Solution Approach 2:
The patent replaces the manual self-reporting mechanism with an automated machine learning-based classification system. Instead of relying on merchants to accurately self-report their category codes, the system uses algorithms to analyze payment activity data and automatically determine the correct merchant classification, substituting human reporting with computational analysis
2Device complexity
If traditional MCC classification is used, then the system is easy to implement and maintain, but the ability to capture contextual information and provide accurate fee structures deteriorates
Solution Approach 1:
The system transitions from the traditional single-dimension MCC classification to a multi-dimensional classification framework that incorporates various contextual features from payment activity data. By analyzing multiple dimensions such as transaction frequency, amount patterns, timing, and merchant behavior, the system creates a richer, more nuanced classification that captures contextual information while maintaining manageable complexity through structured data processing
3Stability of the object's composition
If static merchant classification is used, then the classification structure is stable and predictable, but the ability to adapt to changing business conditions and provide dynamic fee structures deteriorates
Solution Approach 1:
The system implements dynamic merchant classification that can adapt to changing business conditions. By continuously monitoring payment activity data and using machine learning models, the system automatically updates merchant classifications when business patterns change, enabling flexible fee structures that reflect current business operations while maintaining overall system stability through controlled update mechanisms
Data Source
AI summary
Techniques and arrangements for industry vertical classification of merchants using merchant signals, based, in part, on data obtained from payment activity. The techniques can include identifying one or more clusters using the data associated with the merchant and classifying and/or reclassifying the merchant within a cluster and then a class using the one or more identified clusters.