Sales Volume Decomposition Using ML Driver Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for sales volume decomposition of products with high fluctuations in sales drivers, such as perishable items, underestimate the impact of short-term price changes and fail to separate the volume contributions from sales drivers like store location and substitute products, leading to inaccurate base volume calculations.
Innovation Solution
A processor-implemented method and system that integrates transaction-level historical sales data with predefined sales drivers to compute reference values and estimate sales volume contributions using a trained machine learning model, separating the effects of sales drivers like price, competitor price, and weather to derive accurate reference base volumes and recommend real-time retail strategies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current systems use non-promoted sales period to calculate base volume, then base volume can be derived, but price effect is partially included and additional volume associated to price is underestimated
Solution Approach 1:
The patent segments base volume calculation into two distinct components: (1) volume from non-promoted periods without price changes, and (2) volume from non-promoted periods with price changes. This segmentation allows separate measurement of price effects and non-price factors, resolving the contradiction by preventing price effect contamination in base volume while maintaining calculability.
Solution Approach 2:
The patent extracts price effect from base volume calculation by identifying and isolating transactions occurring during non-promoted periods with price changes. This extracted price effect is then separately analyzed, allowing base volume to represent only non-price driven sales while capturing price sensitivity in a dedicated metric.
2Measurement precision
If current systems include store location, space allocated and substitute products in base volume, then base volume reflects overall sales, but volume contributions from these sales drivers cannot be separated
Solution Approach 1:
The patent segments volume analysis into base volume (representing organic demand) and additional volumes (representing sales driver impacts). By maintaining this segmentation, the system preserves both the overall base performance metric and the ability to analyze individual sales driver contributions separately, preventing information loss.
Solution Approach 2:
The patent introduces an intermediary analytical layer that models the relationship between base volume and sales drivers. This intermediary layer uses machine learning to decompose observed volume changes into contributions from base demand and various sales drivers, enabling separation of effects while maintaining accurate base volume representation.
3Loss of information
If retailers want complete visibility on sales drivers, then detailed decomposition is needed, but system complexity increases
Solution Approach 1:
The patent implements self-service through automated machine learning models that automatically decompose sales volume into driver contributions without requiring manual configuration or complex analytical setups. The system self-adjusts to different products, stores, and time periods, reducing operational complexity while maintaining comprehensive sales driver visibility.
Solution Approach 2:
The patent uses parameter changes in machine learning models to adapt the decomposition analysis to different contexts (products, stores, time periods). By changing model parameters rather than restructuring the entire system, the patent achieves comprehensive visibility with manageable complexity through flexible, context-aware analysis.
Data Source
AI summary
Most current systems decompose sales volume of a product into base volume and additional volumes associated to each sales driver. Here base volume refers to sales quantity that is derived from non-promoted sales period. It leads base volume of a product with partial price effect. In addition, base volume includes volume contributed by store location, space allocated to product category and assortment etc. In the present disclosure a system and method for sales decomposition is described by computing reference base volume and sales contribution by each sales driver. Here reference base volume is the volume derived by excluding the effect of sales drivers namely, retailer price, competitor price, demographics, weather conditions, space occupied by the category to which the product belongs, number of available substitute products in assortment. Retail strategies are recommended in real time based on sales contribution of the sales driver for the product.