Fall has just begun, but for companies in the Halloween candy and costume business, spooky season kicked off this summer.
At Mars — the maker of candies like M&Ms, Twix, and Skittles — they’ve found that consumers want to maximize their Halloween festivities, Ramesh Kollepara, the SVP and global chief technology officer of the company’s snacking division, told Business Insider. They’re leaning into “Summerween” by buying trick-or-treat candy in July, Kollepara said, pointing to Mars’ recent in-house report on consumer candy trends.
“That means we have to think beyond a traditional Halloween selling window,” Kollepara said. “We need to understand not only how much demand we’ll see, but when it will begin, where it will occur, and how the product mix may change.”
For this reason, Kollepara called Halloween “a great stress test” for supply chains. There’s added complexity because the Halloween shopping season has a fixed deadline for merchandise to reach stores, said Benjamin Bond, SVP of client strategy and success at Simbe, a retail robotics and AI company. At the same time, ordering too much stock leads to waste.
For manufacturers like Mars and retailers like Walmart, AI helps find the sweet spot. Previously, supply chain planners manually adjusted parameters, and forecasts were trial and error, said Madhav Durbha, the group VP of manufacturing industry strategy at AI platform RELEX Solutions. Now AI automates changes, so planners aren’t “constantly fiddling with the algorithms,” he said.
Using AI to find inventory sweet spots
Mars has turned to AI to track and plan when its treats are in high demand. AI layers in data showing consumers are buying earlier, as well as Mars’ available inventory, production capacity, and customer requirements, Kollepara said.
At Mars, AI connects the dots among four key questions, with results that can vary by geographic region and consumer behavior, Kollepara said. Those questions are:
- What will consumers want?
- What should Mars make?
- Where should the inventory go?
- When will they want it?
For instance, Mars’ research found that consumers in the Western US want Halloween-themed novelty treats, while those in the South prefer gummy candy, allowing Mars to route its inventory accordingly.
Like Mars, Walmart’s use of AI in the supply chain begins with forecasting, Indira Uppuluri, the retailer’s SVP of supply chain technology, told Business Insider. Walmart pulls last year’s Halloween data, new candy releases, and social media trends, then analyzes the data using AI to inform the orders it places with suppliers, Uppuluri said.
Walmart monitors its distribution networks with AI to ensure products are packed and transported in time for trick-or-treaters. Its AI reviews what’s in each trailer and compares it with the current store inventory. The system then flags which trailers to unload right away, and which can wait — a useful tool during the holiday rush.
“At the end of the day, it’s the flow that’s super important for us,” Uppuluri said. “There’s always something coming in; there’s always something going out. How do we get the right movement for the right inventory?”
Executing during the Halloween season
Getting products to warehouses is only part of the Halloween supply chain puzzle, Bond said.
Retailers must also move inventory from stores’ backrooms to shelves and respond to shifts in demand at speed. “Stores have roughly a 10-day window around peak demand to get the season right,” Bond said.
During the season, if Walmart’s AI recognizes signs of too much or too little inventory, “we course correct along the way,” Uppuluri said. It might tell a vendor to pause on sending an order if there’s too much unsold candy.
As candy and costumes arrive at Walmart’s distribution centers, automation technology sorts the goods into pallets based on where the merchandise will go in store aisles, Uppuluri said. That way, store associates can quickly unload goods and stock shelves.
Walmart builds seasonal merchandise pallets early and holds them at distribution centers to avoid crowding store backrooms, which have more limited space than warehouses. When a store needs inventory, a pallet is ready to go. The pallets also have IoT tags and RFID tags, so Walmart can ensure they’re dropped off at the right store.
For all the complex algorithms running in the background, the indicator of success is simple: whether a consumer can get the candy and costumes they want, Uppuluri said.
He added: “You can have smart systems, but if you don’t make that assortment available to customers when they need it, nothing else matters.”
Read the full article here
