AI-driven analysis of purchasing motivations for product assortment decisions in retail
Vaurentis DemandVector
Project description
Today, retailers use a wide variety of systems to prepare decisions regarding product assortments, prices, and promotions. However, the reasons behind customers’ decisions are often only understandable in hindsight. The project therefore takes a new approach that consistently examines retailers’ proactive product assortment decisions from the customer’s perspective. The goal is to better understand and evaluate the demand potential of products as well as the composition of product assortments based on underlying purchase motives, and to use this insight to optimize product placements.
Six fundamental purchase motives, which together describe a “motive space,” serve as the starting point for informed decisions. This allows for the systematic identification and analysis of overlaps and discrepancies between customer needs on the one hand and visual and textual product characteristics on the other.
The goal of the project is to develop a prototype that performs this mapping through data extraction and processing, as well as through the use of rule-based and AI-driven models.
Research contribution
fortiss is researching suitable approaches and workflows for the robust extraction of information from textual and visual product data. To this end, the company employs both traditional rule-based methods and modern AI techniques.
The AI methods focus on language models and image-processing foundation models that can identify relevant concepts from product descriptions and product images and map them to standardized categories. The information obtained is used to classify products according to the six fundamental purchase motives, thereby laying the groundwork for data-driven decisions in retail.
Partner
Vaurentis IP GmbH
Project duration
15.09.2026 - 30.11.2026
