The requirements for retail success are no more fundamental than the ability to accurately forecast customer demand. Even a family-owned bodega should have a pretty good idea of how many people will order a breakfast sandwich and coffee each day. And for larger-scale retailers, this need for accurate forecasting becomes complicated, and further complicated, by the need to properly place and price inventory across multiple locations in order to achieve maximum sales with minimum discounts.
Can artificial intelligence (AI) give retailers and their supply chain partners the guidance they need to improve demand forecasting and inventory optimization? It’s already happening and it’s driving measurable results across a range of use cases. One of the most promising challenges is one that has plagued merchandisers, marketers, and inventory managers for decades, even centuries: demand forecasting.
“Demand is typically the most important input into a company’s operations,” said Rupal Deshmukh, partner in the Strategic Operations group at Kearney in an interview with Retail Touchpoints. “Poor demand forecast accuracy is equivalent to cash. For example, in the chemical industry, if you are at 40% to 50% forecast accuracy, you probably want two to three times more inventory than your peers.»
And therein lies the opportunity.
Learn more about how retailers are driving new efficiencies in their inventory practices with AI and see it in action on Floor and decoration in this free special report Retail TouchPoints.
Unlock the power of unstructured data
Even when a retailer’s overall demand forecast is optimal, inventory placement presents a significant challenge. . Retailers are well aware that having too much inventory in the wrong store, or not enough in the right store, not only affects sales, but also labor costs and store operations. These mismatches also create the need for discounts or additional shipping costs to get products to where they are most likely to be sold.
Compared to demand forecasting, “inventory is more of a difficult math problem to solve,” Deshmukh said. “The equation is simple: there is so much variability in demand, so much safety stock will you need, but inputs are often wrong, and that’s where you get ripple effects.
Deshmukh noted that inaccuracies in retailers’ forecasts often come from an over-reliance on their own internal data: “They often have a lot of point-of-sale data, but they don’t use (the data that) the product companies have or even the data that consumers have,” she said. “They don’t listen to consumers or market signals; we keep seeing that they are not doing things correctly. Companies that are doing well have real momentum on market signalsfor example what consumers see on social networks mediaas well as weather trends and the impact of geopolitical events.”
This is an area where AI can shine, she explained: “AI allows you to evaluate unstructured data in a much more structured way,” Deshmukh said. “The use of large language models (LLM) and generative AI can be useful in the area of customer contributions. Many companies find that their sales teams receive information through the emails and calls they receive, but they cannot convert that information into data points. The AI generation could be a game changer in this area.
AI’s ability to collect and synthesize data from many sources also makes it valuable for retailers managing another part of their inventory: ad placements in retail media. By better understanding not only the number of shoppers in a store (or on a website) at any given time, but also their demographics, lifetime value, and purchase intent, retailers can both improve advertiser campaign performance and quantify incremental benefits.
True progress requires humans who understand the potential of AI
Many of AI’s improvements in demand forecasting and inventory optimization are already recognized, but industry experts agree that there is still plenty of headroom for additional benefits. However, harnessing these benefits will require more than just technological advancements; the mentality of users must also change.
“We’re still missing people who have the vision to understand what’s possible (with AI) and the ability to connect that vision to people who can ask the right questions,” said Fabrizio Fantini, PhD, vice president of product strategy at Tool group in an interview with Retail Touchpoints. “The first feedback we get is that before working with these AI tools, we didn’t even know what was possible. Well, now that you know it is possible, you must acquire the ability to systematically obtain these (tools) and integrate them into daily operations.”
When this happens, it opens opportunities for a highly integrated set of customer-focused systems. “Ultimately, any type of data concerns the customer, directly or indirectly,” Fantini said. “When you suddenly have the ability to systematically and cost-effectively integrate data into your decisions, your ability to create an outcome (that you want) will be much more integrated. Supply chain decisions will be integrated with pricing and marketing decisions as well as financial and budgetary decisions; it’s part of a continuum.”
Retailers need to tap into “the pockets of goodness that AI is creating in a lot of businesses,” Deshmukh added. “We have seen AI applications for demand sensing drive tremendous value; telecom and home goods retailers using external market signals to increase forecast accuracy have led to improvements of up to 10 to 20 percentage points. Additionally, companies that have started using AI in their sourcing and procurement spaces, to test categories or quantify huge amounts of spend data, are doing so faster and with greater accuracy.
Fantini believes that the effects of AI in this area are becoming even more prevalent as technology advances. “These AI models unlock the capabilities of supply chains to do good, and everyone wins,” he said. “The consumer gets better service at potentially lower costs, the business gets more efficiency and profits. They reduce waste, improve environmental problems and improve economic performance by improving the functioning of the market.”
“The supply chain is something that you only know exists when it doesn’t work, but (it can fail) in different ways.“, Fantini added. “However, if you make it more efficient, everyone wins.”
Learn more about how retailers are driving new efficiencies in their inventory practices with AI and see it in action on Floor and decoration in this free special report Retail TouchPoints.