Neural Network Quote Approval Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Configure-Price-Quote (CPQ) systems lack guidance for sales users on the approval likelihood and time required for quotes, leading to uncertainty and inefficiency in the approval process, especially for complex configurable products.
Innovation Solution
Integration of a machine learning component using neural networks to predict approval likelihood and time, providing recommendations on attribute changes to enhance approval chances and reduce approval time, by analyzing historical quote data and attribute interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If neural network models are integrated to predict approval likelihood and time, then user knowledge and decision-making capability are improved, but system complexity increases
Solution Approach 1:
The patent introduces neural network models as intermediary components between the quote submission system and the approval process. These models act as mediators that analyze quote attributes and predict approval outcomes, providing users with actionable insights without requiring them to understand the complex underlying approval workflows. The models serve as an information bridge that translates complex approval dynamics into simple probability scores and time estimates.
Solution Approach 2:
The system enables self-service by providing users with automated predictions and recommendations. Instead of requiring users to manually track approval status or consult with approvers, the neural network models automatically analyze quote attributes and provide real-time feedback on approval likelihood and expected processing time. This allows users to independently optimize their quotes before submission.
2Productivity
If gradient-based attribute recommendations are provided to optimize approval chances, then approval efficiency is improved, but computational requirements increase
Solution Approach 1:
The patent utilizes gradient-based parameter optimization to recommend attribute changes. By calculating gradients of the neural network predictions with respect to quote attributes, the system identifies which parameter changes will most significantly improve approval likelihood. This approach transforms the complex problem of quote optimization into a series of simple parameter adjustments, enabling efficient recommendations without exhaustive search.
Solution Approach 2:
The system provides targeted recommendations for the most impactful attributes rather than optimizing all quote attributes equally. By focusing computational resources on the top gradient-magnitude attributes, the system achieves significant approval efficiency improvements with minimal computational overhead, rather than performing exhaustive optimization across all parameters.
3Measurement precision
If multiple neural network models are used to predict both approval likelihood and time, then predictive accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the prediction task into two specialized neural network models: one dedicated to predicting approval likelihood and another for predicting approval time. This segmentation allows each model to be optimized for its specific prediction target, improving overall accuracy. The modular architecture enables independent training and inference for each model, reducing the computational burden compared to a single monolithic model attempting to predict all outcomes simultaneously.
Solution Approach 2:
The system performs predictions in advance of the actual approval process. By calculating approval likelihood and time estimates before quote submission, users can make informed decisions about quote optimization. The preliminary predictions are based on historical data and trained models, providing accurate forecasts without requiring real-time computation during the approval workflow.
Data Source
AI summary
Embodiments operate a configurator that generates a quote for a product or service configuration, the quote including a quote workflow that includes at least one required quote approval and the quote includes a plurality of quote attributes that define the quote. Embodiments input the plurality of quote attributes into a first neural network model and a second neural network model and generate with a gradient descent a likelihood that the quote will be approved and a time required for the quote to be approved. Embodiments generate one or more attributes with the largest gradient values for the likelihood that the quote will be approved and the time required for the quote to be approved. Embodiments receive a change to one or more of the attributes and regenerate the likelihood that the quote will be approved and/or the time required for the quote to be approved based on the change.