Asahi Breweries and NEC are working on achieving more accurate sales forecasts and the optimization of inventory levels utilizing AI technology and big data analysis technology. What were the circumstances that led to the adoption, what are the current issues, and how do you see things expanding in the future? We spoke with Deputy General Manager Kaoru Yamamoto of the Digital Strategy Department who is leading this project and Shizuka Kanie of NEC.
Adopting AI technology to forecast demand for new products
Starting in September 2014, Asahi Breweries adopted NEC’s AI technology for heterogeneous mixture learning. Can you tell a little about this?
That’s right. We are working on forecasting demand for new products based on multiple data, such as product delivery and actual sales data for the past few years, and calendar and weather data. In the past, we relied heavily on the experience and intuition of experienced personnel, but this was not the type of know-how that anyone could easily pick up. We considered the adoption of AI to make it possible for personnel with limited experience to make forecasts, enable rapid decision-making, and optimize product plans and inventory.
NEC and Asahi Breweries have a long-standing relationship through the provision of their core system and other services. However, NEC has world-class technology and data scientists that are active in the distribution and retail industry, so we believe that we can assist more directly on the operation side.
What phase is the project for forecasting demand for new products currently in?
We had achieved a certain level of accuracy in the verification phase that started in September 2014, and this has improved further over the past three years. We are currently continuing verifications while also working on the operation design in preparation for full-scale adoption in the worksites.
Aiming to provide proposals for more optimal sales promotion plans for the retail chain
Additionally, starting December 2014, the project for demand forecasting and price optimization for the retail chain will also start. What kind of issues were there?
The retail industry continues to experience structural recession with the declining and aging population, increase in consumption tax and heightened competition from the expansion of the EC business, and there was the issue that sales in the retail chain were not growing as expected through conventional sales promotion and measures. NEC was feeling the need to strengthen our ability to propose factors other than sales prices, such as by holding seasonal events at appropriate moments and stirring up the consumption of experiences. That was when we received this proposal that perhaps the knowledge gained from the project for forecasting new product demand could be useful in resolving issues in the area of retail, and so we have started our efforts.
Using highly accurate forecasts based on NEC’s big data analysis, it is now possible to make proposals for optimal price sales promotion and measures to retail chains that are more persuasive. The team has conducted various discussions to give shape to ideas about how this can be used to achieve results.
During the past three years working on these two projects, were there any hurdles you had to overcome?
NEC has superior technology, but to adopt the technology requires some degree of literacy on the part of the user. At the time, in our company, there were only a few people who properly understood the meanings of “AI” and “Big Data.” So, it was necessary to start by having project members speak using a common language and to mutually compensate for any insufficiencies in their knowledge.
NEC receives high assessment from our customers regarding our technology, but that does not mean it will completely fit the operations of each individual customer. In addition to helping them adopt the technology, we must understand the issues with the customer’s operations, and we rack our brains every day trying to make improvements. Things do not always work out right away, so I am experiencing for myself the difficulties everyone in the operations departments face in utilizing the technology, and the entire team is working together to take on this challenge.
I believe that the key to the future of this project is seeing how far we can maximize value by combining our knowledge. The project team has now grown to about 20 members, and I believe we have formed a good team that can stimulate each other in a positive way and grow together. We are slowly starting to see results at the worksite, but we hope to develop this much further.
It is people who utilize technology. We are aiming for further optimization.
Please tell me about your future plans.
We tend to have excessive expectations that technology will solve any issue, but in many cases, the only way to obtain maximum results is to review operations and change people’s awareness instead of just adopting the technology. To maximize value, we hope that everyone at NEC will continue to give us their frank opinions and proposals as part of this team. We are still working on individual optimization, but in the future, we hope to expand the utilization of AI technology to the entire Group from the perspective of overall optimization.
The project for forecasting demand for new products with the aim of optimizing inventory levels and the project for demand forecasting and price optimization aiming to contribute to the increase in sales in the retail chain are currently underway. In the future, we will link and develop these two projects so that NEC’s technology and knowledge can contribute to business expansion for the entire Asahi Breweries Group.