en.Wedoany.com Reported - On June 8, Toshiba announced the development of a sales forecasting AI technology based on purchase big data from its electronic receipt service "Smart Receipt®." This technology can be used to predict sales volumes for new product launches, targeting consumer goods sectors such as food and daily necessities. Toshiba plans to expand the scope of verification and aims to commercialize the service by the fiscal year 2027.

The core of this technology lies in converting real consumption data from electronic receipts into product sales forecasting capabilities. Toshiba stated that traditional sales forecasting often suffers from deviations when faced with diverse consumer preferences, accelerating market changes, and a lack of historical sales data for new products. Directly modeling each consumer individually would cause computational costs to rise rapidly with user scale. The newly developed sales forecasting AI employs a combination of two types of AI: one is a clustering AI developed for purchase big data, which automatically groups consumers with similar purchasing behaviors; the other is a generative AI that calculates a "reaction score" for new products based on group characteristics, which is then combined with the size of each group to estimate overall market sales volume. By grouping first and then forecasting, the system can reduce computational burden while incorporating preference differences among various consumer groups into the model.

Toshiba disclosed that "Smart Receipt®" has over 3 million members. Evaluation experiments based on purchase big data from this service showed that, in sales forecasting for specific product categories, the new technology can reduce the error between predicted sales volume and actual sales volume by approximately 23% compared to traditional forecasting methods.

For retail and consumer goods companies, sales forecasting directly impacts new product development, production planning, channel distribution, inventory control, and promotional pacing. In the past, small and medium-sized brands and local enterprises often lacked sufficient data and algorithm teams, making it difficult to build sophisticated demand forecasting systems like large consumer goods companies. The advantage of electronic receipt data lies in its coverage of actual purchasing behavior, which can be continuously updated as consumption changes. AI models can identify preference shifts, category substitutions, seasonal factors, and consumer group differences from vast amounts of historical purchase records. By applying this capability to new product sales forecasting, Toshiba can help companies assess potential demand earlier before product launch, reducing risks of overproduction, stockouts, and unsold inventory, while also providing a basis closer to market changes for product development, pricing, and inventory allocation.

This technology also reflects that the value of electronic receipt services is extending from "paperless vouchers" to data intelligence. Paper receipts only serve as transaction records, while electronic receipts, under user authorization and compliant usage, can accumulate continuous, structured purchase data, enabling new applications such as consumption trend analysis, product forecasting, retail media, and supply chain optimization. If Toshiba subsequently integrates this technology into its purchase big data analysis service and opens it to manufacturers and retail-related companies, it will help connect consumption data, AI forecasting, and supply chain management. Details of this technology will be presented at the JSAI2026 National Conference of the Artificial Intelligence Society, held from June 8 to 12, 2026. Whether it can evolve into a large-scale service will depend on factors such as expansion of forecasted categories, data authorization mechanisms, customer system integration, and accessibility for small and medium-sized enterprises.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com