Better Accuracy, Fewer Stock-Outs, Happier Customers: How Six Companies Use AI For Demand Planning

Feature

carol millerRetailers, manufacturers and distributors are increasingly leveraging Artificial Intelligence-powered software to aggregate and analyze data for forecasting, allocation and replenishment planning.

If you’ve ever been frustrated because your favorite brand of ketchup is sold-out at the grocery store or had to postpone a home improvement project because some of the key materials were out-of-stock in your region, you’re not alone. Retailers, manufacturers and distributors all share your aggravation. Indeed, they’re likely even more frustrated, considering they have teams dedicated to forecasting, allocation and replenishment planning.

Historically, these processes have all been manual, time-consuming, error prone and reactive—instead of proactive. Planners making phone calls to suppliers and stores, inputting data into spreadsheets, looking at what happened in the past, analyzing statistics and relying on intuition was the way of this world for decades. But with the increasing availability of affordable cloud-based computing power, data aggregators and advanced artificial intelligence (AI) algorithms and machine learning (ML) technologies, the world of demand planning has experienced a seismic shift.

McKinsey researchers found that AI-enhanced supply chain management significantly improves forecasting accuracy while simultaneously increasing granularity and optimizing stock replenishment. The firm notes AI can cut forecasting errors by 20% to 50%. This boost in predictive accuracy can cut lost sales due to stock-outs by up to 65%, while safety stock inventory can be reduced by as much as 50%.

Where applying AI to demand planning truly adds value is in the ability to factor in external influencers that impact demand, said Knut Alicke, a partner in McKinsey’s Stuttgart, Germany office.

“Think about weekends in the summer, when barbecue and beer consumption depend a lot on weather,” he explained. “Sunny weekends will have higher consumption than rainy weekends. AI-powered forecasting pulls in this information, whereas the old forecasting approach would have looked at the same weekend from the previous year, when the temperature may have been much lower, and sales were down.”

While AI can certainly help with demand planning, Alicke cautioned against companies deploying it in a silo. “There’s been an increase in interest in using AI for planning because companies are eager to make accurate predictions and respond more effectively in the face of the uncertainty over the last two-and-a-half years,” he noted, referring to the pandemic. “While there’s talk of autonomous planning, you always need an experienced planner to evaluate the algorithm’s predictions.”

Alicke believes demand for 80% of a company’s products can be forecast by AI with no human assessment, saving planning teams time. As for the other 20%, there are always exceptions. “When people rushed to buy toilet paper in 2020 during the lockdowns, an AI algorithm would likely project that increase to continue. But an experienced planner would recognize that actual toilet paper consumption would likely remain steady and predict buying patterns to return to normal—which they did,” he said.

One of the biggest advantages of AI in demand planning is the ability to prevent stock-outs, Alicke added. He noted that if a product is not available, then customers will either buy it from a competitor or not buy it at all. That results in a loss of revenue. “With AI-driven planning, there’s much more accuracy in product availability. That correlates to an increase in revenue and margin without significantly increasing the amount of inventory,” he concluded.

Better accuracy and fewer stock-outs equate to happier customers. That’s why retailers, manufacturers and distributors are increasingly leveraging AI to aggregate and analyze data for forecasting, allocation and replenishment planning. Here’s how six companies are doing just that.

PacSun boosts inventory efficiency and lowers costs through AI-driven allocation and fulfillment

pacsun implemented

PacSun implemented a new omnichannel strategy to reduce split shipments and order zone hopping for online fulfillment. The company partnered with antuit.ai to leverage its Allocation AI solution. PHOTO PROVIDED BY PACIFIC SUNWEAR OF CALIFORNIA

Operating nearly 400 stores, Pacific Sunwear of California (PacSun) is a top three Gen-Z clothing brand. After the 2019 holiday season, but before the pandemic, the retailer recognized the need to adjust its operations to account for online growth. Because its DCs didn’t have the necessary capacity, PacSun stores had to fulfill orders, often in partial shipments. Further, shipping costs were rising rapidly. COVID-19 exacerbated all of these factors.

“PacSun wanted to be ready for the next holiday season by proactively managing our costs created by the increase in online sales,” said Mike Relich, PacSun Co-CEO.

To lower shipping costs before the 2020 holiday season, the company opted to implement a new omnichannel strategy that aimed to reduce split shipments and order zone hopping for online fulfillment. Rather than every store fulfilling online orders, they concentrated online orders into specific locations called web-depot stores. The goal was to balance inventory across DCs and stores by having web-depot stores handle approximately 30% of peak volume within two shipping zones.

For the new strategy to work, inventory had to be properly allocated across DCs and stores to manage both online and store demand. PacSun partnered with antuit.ai, a part of MHI member Zebra Technologies, to leverage its Allocation AI solution.

PacSun designated 100 web-depot stores based on their location and footprint. Each serves specific zip codes for online fulfillment. Demand forecasting from antuit.ai’s Allocation solution considers both in-store and online demand to concentrate the optimal amount of inventory in these stores.

“By adding more intelligence to our allocation processes with antuit.ai, we allocated inventory for the omni-channel demand and minimized split shipments,” Relich continued, adding that PacSun calculated that just a 1% reduction in shipping costs would pay for the project.

Additionally, the company doubled its ship complete rates and reduced overall costs, agreed PacSun CIO Shirley Gao. “antuit.ai’s solutions and expertise ensure we have the right inventory in the right stores, which significantly improves our ship completes,” she said.

AI/ML-driven forecasting delivers accurate demand insights for Canadian tech retailer The Source

the source canada

The Source, Canada’s largest tech retailer, integrated data-driven forecasting into its supply chain planning process by partnering with Blue Yonder and using the company’s demand planning solution. PHOTO PROVIDED BY THE SOURCE

The Source, Canada’s largest tech retailer with locations in most major malls across Canada, sells a broad selection of consumer electronics. Through more than 300 stores and online, consumers can find smartphones, gaming consoles and accessories, headphones, home entertainment, laptops and more. Additionally, The Source’s inventory moves at a range of throughput velocities, including low frequency electronics, such as TVs; hot, in-demand offerings, such as video game consoles; and fast-moving conveniences, such as chargers. As a result, accurately forecasting demand for such a wide range of products is often challenging.

To gain greater precision in inventory allocation and distribution across all its channels, The Source sought to integrate data-driven forecasting into its supply chain planning process. The retailer selected MHI member Blue Yonder’s demand planning solution, part of the Luminate Planning portfolio.

Powered by Microsoft Azure, the Blue Yonder solution utilizes AI and ML to enable The Source to better plan for current and future consumer demand of products to meet changing buying patterns. The AI and ML capabilities allow The Source to incorporate hundreds of variables that drive demand. In turn, the company receives a single, unbiased and automated demand forecast that highlights calculated business risk and impact.

Managers can then evaluate scenarios and leverage prescriptive recommendations to enable more accurate decisions; from staging the right inventory through the distribution network to minimizing stock-outs while maximizing inventory turns. The Blue Yonder demand planning solution also enables The Source to quickly identify and reduce unproductive inventory in its stores and DCs, as well as improve promotional sales in its stores.

“The consumer electronics sector has experienced unprecedented challenges with supply and demand,” said Declan Brady, vice president of merchandising and supply chain at The Source. “Blue Yonder’s solution has enabled us to optimize areas of opportunity and share forecast data with our vendor community, which in turn empowers us to provide an exceptional omnichannel customer experience while maintaining a healthy supply chain.”

MacPherson’s reduces uncertainty around import lead times with AI-powered supply chain planning solution

MacPherson’s is the largest art supplies distributor in North America. The company serves 2,600 retail customers across a variety of channels, including independent mom/pops, regional chains, mass market and e-commerce. Their vast product portfolio spans more than 50,000 products and 250 brands sourced from vendors worldwide.

With a large share of the company’s assets going toward inventory, MacPherson’s leadership wanted technology that could reduce costly uncertainties in planning. A key priority was to preserve the company’s high standard for service. But the expensive alternative of carrying a high volume of safety stock to make this happen was not ideal. In addition, MacPherson’s struggled to get a handle on extremely long lead times for Asian imported products.

A veteran user of Blue Ridge Global’s inventory optimization offering, MacPherson’s subsequently added the company’s cloud-native SCP supply chain planning solution that integrates AI, ML and data modeling. This enables the distributor to further refine inventory levels and automate their demand planning with greater precision and agility.

Blue Ridge SCP’s cloud-native architecture allows MacPherson’s to abandon their previous buying method: spreadsheets. That approach often resulted in inaccurate forecasting, imprecise recommendations and slow reaction to shifting trends, disruptions and conditions. Rather than deploying averages, estimates and assumptions, MacPherson’s inventory analysts now use data-driven intelligence, gaining both greater visibility and a realistic, timely picture of product availability.

By broadening demand and supply planning projections—while still being able to access granular details—the solution lets MacPherson’s analysts make informed buying decisions that protect their customer commitments. One of the biggest areas where this precision particularly helps MacPherson’s is their Asian import buying, where lead times are extremely long. Further, since implementing Blue Ridge SCP, MacPherson’s has improved fill rates to 95%, meeting their company service goal.

Click here to read the full feature.