Merchant-Specific Fraud Thresholds Using Dual Machine Learning Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing fraud detection systems in online transactions struggle with setting customized fraud tolerance thresholds, as merchants have varying risk tolerances and may not fully understand the consequences of their threshold settings, leading to compromised security and potential growth impediments.

Innovation Solution

A machine learning model determines transaction fraud likelihood and a separate model sets a suitable threshold for each merchant, allowing for customizable fraud tolerance settings based on merchant-specific preferences and historical data, optimizing revenue while reducing fraudulent transactions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single fraud tolerance threshold is used for all merchants, then the system is simple to operate, but it compromises security for merchants with low tolerance and impedes growth for merchants with high tolerance

Engineering Contradiction:

Improvefraud prevention effectivenessVSAvoidthreshold configuration simplicity

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The patent applies local quality by assigning different fraud tolerance thresholds to different merchants based on their individual risk profiles, transaction patterns, and business needs. Each merchant receives a customized threshold rather than a universal one, allowing the system to adapt to local (merchant-specific) requirements while maintaining overall system security and growth balance

Inventive Principle:

Principle #3Local quality

2Reliability

If merchants set custom fraud tolerance thresholds, then security can be optimized for each merchant, but merchants may not understand the consequences leading to suboptimal settings

Engineering Contradiction:

Improvecustomized fraud protectionVSAvoidmerchant understanding of threshold consequences

Core Design Contradiction:

ReliabilityVSDifficulty of detecting and measuring

Solution Approach 1:

The system enables merchants to self-configure their fraud tolerance thresholds through an intuitive interface that presents options with clear, simplified explanations of consequences. Merchants can adjust their preferences without needing to understand complex fraud detection mechanics, as the system translates technical parameters into business-impact language and provides guidance based on their transaction history and risk profile

Inventive Principle:

Principle #25Self-service

3Productivity

If a high fraud tolerance threshold is set, then transaction growth is encouraged, but fraudulent transactions are more likely to slip through

Engineering Contradiction:

Improvetransaction throughputVSAvoidfraudulent transaction risk

Core Design Contradiction:

ProductivityVSObject-affected harmful factors

Solution Approach 1:

The patent implements dynamic threshold adjustment where fraud tolerance levels are not fixed but adapt in real-time based on transaction characteristics, merchant behavior patterns, and detected fraud trends. The system can dynamically raise or lower thresholds for specific merchants or transaction types, allowing high throughput for low-risk transactions while maintaining strict filtering for suspicious activities

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the fraud detection parameters (thresholds) based on multiple factors including transaction amount, frequency, merchant history, and risk indicators. By dynamically adjusting these parameters rather than using static values, the system optimizes the balance between allowing legitimate transactions to proceed and blocking fraudulent ones

Inventive Principle:

Principle #35Parameter changes

4Object-affected harmful factors

If a low fraud tolerance threshold is set, then fraudulent transactions are blocked more effectively, but legitimate transactions may be impeded reducing growth

Engineering Contradiction:

Improvefraudulent transaction blockingVSAvoidlegitimate transaction flow

Core Design Contradiction:

Object-affected harmful factorsVSProductivity

Solution Approach 1:

The patent segments the fraud detection process into multiple layers and channels, applying different threshold levels to different transaction categories, merchant types, and risk profiles. High-value or suspicious transactions receive stricter scrutiny with lower effective thresholds, while routine low-risk transactions experience more lenient thresholds, ensuring that fraud blocking does not unnecessarily impede legitimate business operations

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250272692A1Machine learning for fraud tolerance

Publication Date: 2025.08.28 STRIPE LLC

AI summary

In an example embodiment, a solution is provided wherein a machine learning model is to determine a likelihood that a transaction is fraudulent, but also a separate machine learning model is used to determine a suitable threshold for a merchant. This predicted suitable threshold can either be automatically applied to the merchant, or can be recommended to the merchant (allowing the merchant to accept or reject it).