ATP Value Determination Using ML Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Omnichannel retailers face challenges in determining optimal available-to-promise (ATP) values, as conventional rules-based approaches fail to account for unique business models and goals, leading to tradeoffs between lost sales and customer satisfaction, and are compounded by complexities such as orders from multiple locations and potential cancellations.
Innovation Solution
A computer-implemented method using machine learning models to identify business constraints and goals, determining numerical parameters for an executable optimization model, which generates optimal ATP values by balancing retailer objectives and constraints, including inventory management, order cancellations, and delivery timing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional rules-based approaches are used to determine ATP values, then the determination process is simple and fast, but the accuracy and optimization for specific business goals are insufficient
Solution Approach 1:
The patent introduces an optimization model as an intermediary between the complex retail operations data and the ATP determination. This model acts as a mediator that processes multiple business constraints, goals, and operational parameters to generate optimized ATP values, thereby improving accuracy without directly managing the complexity of all underlying factors
Solution Approach 2:
The system performs self-service by automatically identifying business constraints and goals from operational data, building the optimization model autonomously, and generating ATP recommendations without requiring manual configuration. This self-organizing capability improves accuracy while keeping the user interface simple
2Productivity
If ATP values are optimized for revenue maximization, then lost sales are reduced, but customer satisfaction may deteriorate due to over-promising
Solution Approach 1:
The patent changes the parameters of the ATP determination by incorporating multiple objective functions into the optimization model, including both revenue maximization and customer satisfaction metrics. By adjusting these parameters dynamically based on business priorities, the system balances the trade-off between lost sales and over-promising
Solution Approach 2:
The system dynamically adjusts ATP values based on real-time operational data, order patterns, and business constraints. This dynamic approach allows the system to adapt to changing conditions, optimizing for revenue when capacity is available and prioritizing customer satisfaction when resources are constrained
3Speed
If real-time ATP determination is performed for each inquiry, then responsiveness to customers is improved, but computational resources and time are consumed
Solution Approach 1:
The system performs preliminary actions by pre-processing operational data to identify business constraints and goals, and by pre-building the optimization model structure. This preparation work is done in advance so that when ATP inquiries arrive, the system can quickly execute the model with minimal computational overhead
Solution Approach 2:
The optimization model is designed to be universal and multi-functional, handling multiple product types, order scenarios, and business constraints within a single framework. This universality allows the system to serve multiple purposes without requiring separate computational models for each scenario
Data Source
AI summary
Generation of an available to promise (ATP) value. A method identifies business constraints and goals for an omnichannel retailer based on input quantitative and qualitative data. The method determines numerical parameters for use in an executable optimization model and using a machine learning model. Additionally, the method builds the executable optimization model using at least one of the goals as a respective at least one objective in the optimization model, at least one of the constraints as a respective at least one constraint in the optimization model, and at least one of the numerical parameters as at least one additional parameter in the executable optimization model. In addition, the method executes the executable optimization model and generates an ATP value, and outputs the ATP value to an e-commerce system.