Overview
AB InBev, a global leader in the beer manufacturing industry, has a presence in a wide range of sales channels in the South Korean market, including OOH (out-of-home) and off-trade. The specifics of retail execution in the market are the high diversity of product placement standards on main shelves, refrigerators, additional display locations, and displays, which makes retail audits and calculating sales team KPIs extremely challenging.
Challenge
The company used a large number of placement standards across sales channels. Different methods of KPI calculations were used for each channel and sub-channel, taking into account the specifics of placement, such as accounting for stocks on the upper and lower shelves, accounting for products stacked on top of each other, multipacks with several articles, and distribution of shelves in large refrigerators without clear boundaries between product categories. This made manual audits inaccurate, impractical, and time-consuming.
AB InBev Korea used an automated Image Recognition solution, but it did not meet expectations due to low processing and data exchange speed – about 40 minutes from photo to result. The company was looking for a new Image Recognition solution that could provide accurate and fast image processing while meeting several essential requirements:
- Integration with the Salesforce solution for field employees
- Customizable calculations for different channels and locations
- User-friendly camera interface
- Ability to quickly train the neural network (short machine learning cycle)
Solution
AB InBev Korea selected SoftServe Business System’s AI Shelf Recognition solution among several proposals. The solution was deployed in just 3 months and included training neural networks for the first 200 SKUs and seamless integration with Salesforce Consumer Goods Cloud.
AI Shelf Recognition was deployed with customizable calculations, enabling processing across various channels and placement locations. This customization is achieved through tailored algorithms based on specific standards and key performance indicators for each channel.
Results
The implementation of SSBS’s Image Recognition retail execution solution brought impressive results for AB InBev Korea:
- 95%+ recognition accuracy
- Recognition speed increased to just 11 seconds (compared to the previous 40-minute processing time)
- The Machine Learning Cycle was reduced to 2 weeks compared to the previous 1-month period
By integrating the Image Recognition solution with the SFDC platform used by field teams, AB InBev Korea achieved rapid calculation of key performance indicators both in the Image Recognition system and on the SFDC side.
The successful implementation of SSBS’s Image Recognition solution has allowed AB InBev Korea to streamline retail audits, enhance accuracy, and make data-driven decisions more effectively in the South Korean market.
