User Session Scoring for High-Resolution Product Demand Forecasting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manufacturers face challenges in accurately forecasting product demand due to the 'white car problem' and the complexity of numerous configuration options, leading to inaccurate sales data and subsequent revenue losses.
Innovation Solution
A system and method that collect and analyze user session data from product selection websites to develop master session profiles, assigning scores based on the likelihood of purchasing, which are used to determine product demand with high resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If manufacturers use actual dealer sales data to forecast demand, then they have data to work with, but the data is inaccurate due to the 'white car problem' where buyers purchase available configurations rather than preferred ones
Solution Approach 1:
The system performs preliminary actions by collecting and analyzing user session data before actual purchases occur. It identifies quality user sessions that exhibit purchasing intent patterns, scores these sessions, and uses them to forecast demand in advance, preventing the white car problem from distorting demand signals.
Solution Approach 2:
The system introduces an intermediary layer between actual sales and demand forecasting. Instead of using raw sales data directly, it uses scored user session data as an intermediary metric that better reflects true buyer preferences, thereby improving the accuracy of demand predictions.
2Adaptability or versatility
If manufacturers increase the number of product configuration options, then they can meet diverse customer needs, but forecasting accuracy decreases due to the complexity of numerous options
Solution Approach 1:
The system segments the complex set of product configurations into individual attribute dimensions (e.g., color, trim level, options). It analyzes user session data at this granular level, allowing manufacturers to forecast demand for each configuration attribute independently, thereby maintaining forecasting accuracy even with numerous product options.
Solution Approach 2:
The system changes the parameter of analysis from aggregate product-level forecasting to attribute-level forecasting. By scoring user sessions based on specific configuration selections within sessions, it transforms the complexity of multiple product options into manageable, scorable attributes that can be analyzed independently.
3Productivity
If manufacturers produce products based on forecasted demand, then they can plan production, but they risk significant revenue loss when actual demand differs from forecasts
Solution Approach 1:
The system implements feedback by continuously collecting new user session data, comparing it against established quality session patterns, and updating demand forecasts accordingly. This ongoing feedback loop allows manufacturers to adjust production plans dynamically, reducing the risk of revenue loss from inaccurate forecasts.
Data Source
AI summary
A session quality system collects and evaluates data from multiple user world wide web (“web”) product selection sessions and correlates the collected data with actual product purchases. The system assembles the data into a set of session profiles and assigns a score to each profile that represents a relative likelihood of purchasing the product. The system matches session data collected from future product selection sessions with a master profile. A weighted average of scores based on matched master profiles provides useful product demand information. Furthermore, the system can determine product demand with varying ranges of resolution. For example, the system can capture specific features of each product, such as color, audio system selections, and wheel types. Since these features can be common across multiple products, manufactures can expand their use of the high resolution product demand information.