AI delivers sales, service improvement for Coca Cola vending machines
Ed DeFraine describes how AI improved managing vending machines, flanked at left by Jason Hosking of Hivery, Reza Kasravi of Coca-Cola North America and Matt Robards of Hivery.
The power of artificial intelligence in automated retailing has been proven in the full line beverage vending industry, a business that requires managing product selection and service for thousands of machines across hundreds of miles.
|Jason Hosking of Hivery gives an overview about artificial intelligence.|
Reyes Coca Cola Bottling, an Irvine, California-based Coke bottler serving California and Nevada, presented the results of a two-year AI project that helped the company to significantly improve sales and reduce service costs.
Ed DeFraine, vice president of food service and on-premise for the bottling company, explained the AI project during last week's National Automatic Merchandising Association show at the Las Vegas Convention Center. He was joined by representatives of the company's AI partner in the project, Australia-based Hivery.
DeFraine told attendees that the project marks the first application of AI in the full line beverage vending industry. The project demonstrated AI's uniqueness as a tool to automatically execute intelligent decisions based on extensive, complex data much faster than humans acting on their own.
Jason Hosking, founding director and CEO of Hivery, began the presentation by defining AI as "computer systems able to perform tasks normally requiring human intelligence."
The technology revolution has unleashed more data on the world than humans can access, Hosking said, pointing out that only 12 percent of available data is currently being accessed by businesses.
AI is not a one-time fix for a problem, Hosking noted. Once it is applied to a specific task, however, it learns and masters the task.
More importantly, and contrary to popular belief, Hosking said AI doesn't replace humans. While AI can complete a task faster and better than humans acting alone, it can't match the performance of humans working in partnership with AI.
"AI plus (human) experts can do amazing things," he said. AI plus medical experts have made significant progress in detecting breast cancer early, for instance.
|Hivery cited numerous industries that stand to improve performance by investing in artificial intelligence.|
Opportunity massive for automated retail
Turning the focus back to automated retailing, Hosking said, "the opportunity in the (automated retail) industry is massive."
The Hivery AI solution makes use of robotics, sensors, natural language processing, speech recognition, text-to-speech, computer vision, machine learning, machine reasoning, decision making, deep learning, neural networks, business analytics and data services.
Matt Robards, a Hivery data scientist, took the stage and explained that the company's cloud-based retail software identifies the right product, the right space-to-sales ratio and the right promotional activity for products in every one of Reyes Coca Cola Botting's 19,000-plus vending machines. The Hivery software sits atop the existing vending machine management software.
The AI software has a "build planogram" function that generates a planogram — a visual diagram that details the placement of every product in the machine — in two minutes as opposed to the 20 minutes it took humans using the existing management software. The tool prescribes changes to the planogram on a per machine basis.
The tool also does a "what if" analysis for a product the company might consider placing in a machine. There is no need for the company to actually test the product to determine how it will perform in the field.
"You know what will happen before you actually do it," Robards said.
Vending revenue was declining
Prior to deploying the AI tool, Reyes Coca Cola Botting was designing machine planograms based on a limited set of location characteristics. The drivers filling the machines were given some freedom in making adjustments to the planograms based on their personal knowledge of an account's characteristics.
The company's existing software selected the products and determined when the machines needed service based on the machines' sales histories.
"We knew we could do better than that," DeFraine said.
Reyes' vending revenue was declining before the company launched the Hivery solution in 2016. DeFraine said he wanted a better understanding of what products were being stocked in the machines and which were best sellers.
Reyes learned about Hivery's AI tool from Coca Cola North America, which was testing the AI software in 25 vending machines in Las Vegas. DeFraine visited some of the locations where the AI software was being used.
He saw firsthand how the AI software was making product placement decisions that would be impossible for a human. In one hospital emergency room, for example, the vending machine carried only one facing of the Monster energy drink instead of the usual 15 to 20 facings. The AI took into account that people visiting emergency rooms don't usually buy energy drinks.
Reyes then launched its own test in Sacramento to observe what impact the tool would have on machine sales and service needs.
In one sports and entertainment stadium, the AI tool allocated two rows in the machines to Minute Maid lemonade beverages. This was because the machine was patronized by roadies, who tend to drink a lot of lemonade.
"It was optimizing based on that specific occasion and that specific consumer set," DeFraine said. No beverage driver would have allocated two rows of lemonade in the machine, he said.
Prior to using the AI tool, it was necessary to import data from different sources and use spreadsheets. The AI tool is one self-contained unit that is easy to use and makes more intelligent decisions, faster. He said it takes the tool two minutes to do a "high quality" optimization.
Accurate data required
In order for the AI tool to work, however, it is necessary for the machines to report accurate data, DeFraine said. Reyes had already equipped half of its machines with telemetry devices that report machine data via cellular signal. These machines account for 80 percent of the company's vending sales.
Sales from machines not equipped with telemetry are measured using predictive modeling methods. The predictive modeling yields accurate data for machines with slower sales, DeFraine said.
The existing technology, however, could not take into account as many factors as AI can. Nor could it accurately predict a product's performance, so it was necessary to test new products, which required a certain amount of additional service and management.
The Hivery AI tool was also able to more accurately determine the best times to service the machines, which made restocking more efficient, DeFraine said. By stocking the machines with the right amount of the right products, the machines don't need to be serviced as frequently.
Once the AI tool was deployed companywide, by the end of 2017, it delivered 6 percent additional revenue and 15 percent fewer restocking trips, DeFraine said.
There are currently three planners working with the AI tool on a daily basis, he said.
"Vending analytics has become a way of life," DeFraine said. "Net net, we're seeing fantastic results using this tool. It's really astounding how powerful this is."
The improvement that AI has delivered to beverage vending has implications for other industries as well, Hosking said during the presentation. He presented a slide noting that an investment in AI and human machine collaboration can improve consumer goods revenues by 51 percent and deliver a 10 percent change in employment. Other industries that stand to gain from this collaboration include health, telecommunications, retail, professional services, financial services, chemical land automotive.
Elliot Maras Elliot Maras is the editor of KioskMarketplace.com and FoodTruckOperator.com.