Toshiba Tec Corporation (HQ: Tokyo; President and CEO: Hironobu Nishikori; hereinafter “Toshiba Tec”) announced today the development of an AI solution for optimizing retail promotions. This solution utilizes a "transformer," a pivotal generative AI engine. By analyzing POS data, it creates an AI model of customer preferences and responses, simulating revenues and profits from various promotional combinations tailored to individual customers. Additionally, to maximize profits, it addresses mathematical optimization challenges, ensuring optimal distribution of coupons and points.

With the widespread adoption of artificial intelligence and machine learning in recent years, personalized promotions leveraging past purchase and behavior data have become commonplace in the retail sector. However, existing methods often struggle to optimize promotions for consumers with limited purchase history. Consequently, these methods typically focus on suggesting top-selling or previously purchased products, leaving newer or less popular items overlooked.
To address these challenges, Toshiba Tec has integrated a "transformer" module into its recommendation model. This allows for the integration of individual consumer purchase histories along with attributes of customers, products, and retail outlets to predict consumer preferences. Furthermore, by analyzing numerous customer-product combinations, it can calculate the impact of coupon distribution on profit and determine combinations that maximize overall profits.

This solution harnesses the extensive AI expertise cultivated by the Toshiba Group over many years, particularly in optimizing supply and demand for social infrastructures. Combined with Toshiba Tec's deep understanding of the retail industry, it aims to enhance resource utilization and measure the impact of sales promotion activities. Additionally, it will generate customer response categories to facilitate customer orientation analysis and personalized solutions, thereby laying the foundation for comprehensive customer orientation visualization.

Toshiba Tec is enhancing the development of services utilizing generative AI, and this solution marks the initial phase of that endeavor. Going forward, we will persist in developing AI solutions to address challenges in the retail sector.

Toshiba Tec's initiatives with generative AI will be showcased at the Toshiba Tec booth during RETAILTECH JAPAN 2024, taking place at Tokyo Big Sight from March 12 to 15 this year.

As we strive to become "A global top solutions partner," we are committed to collaborating closely with our partners to embody our corporate philosophy of "Create with You" This entails promoting digital transformations (DX) aimed at addressing challenges encountered by customers, as well as the logistics and retail industries, thereby contributing to society as a whole.

Component elements of the promotion optimization solution

Encompasses four key functions, integrating the transformer module into function 1. mentioned below.

1. Predict which products individual consumers are most likely to purchase by analyzing combinations of consumers and all available products.
2. Forecast the response level of individual consumers to each promotion for every product.
3. Model fluctuations in revenues and profits resulting from the distribution of each promotion to consumers.
4. Given constraints on promotional investments, analyze vast combinations to determine which promotions targeting which consumers will yield the highest economic efficiency and maximize total profits, then distribute those promotions accordingly.