Why AI-powered smart stores are a smart bet for retailers
John Spencer John Spencer
September 18 9 min

Why AI-powered smart stores are a smart bet for retailers

Today, there's no doubt the customer wins⁠—but does the store? With the right inventory assortment, yes!

Retailers today face waves of disruptive trends. From tariffs to natural disasters and other global disruptions, volatility is impacting supply chains all the way down to product assortment at the store level. Uncertainty around shopping behaviors is the new normal, whether the shopping takes place in the digital realm or the physical one. The complexity of being expected to provide and deliver products within a moment’s notice, to any location, with seemingly endless options, creates massive ambiguity around what belongs in the store, when, where and at what price to keep customers coming back.

Volatility, uncertainty, complexity and ambiguity: these create the “VUCA” world in which we live—and it’s coming to a store near you. In fact, as many of the retailers we’ve spoken with have told us, it’s already here. Take inventory distortion and out-of-stock (OOS) merchandise, which currently cost retailers up to $1 trillion annually. As retailers seek to match the options available to consumers online, products are proliferating across the supply chain, resulting in costly surpluses at the store, warehouse or distribution center (DC). But of course retailers’ motivations are totally rational: they know that if customers can’t get what they’re looking for and are left unsatisfied, they will simply turn to online giants or some other fast and convenient option.

Given the competitive landscape, it’s imperative that stores better meet the demands of today’s consumers: the space, product and price must be exactly what the customer wants. Three-quarters of consumers say that a good customer experience drives their brand loyalty, according to one recent study—and inventory assortment is a critical part of delivering this kind of exceptional customer experience. It’s a tall order, but the good news is that it will pay dividends to retailers that get it right.

So how do we get there, and why is AI poised to solve this problem better than a traditional planning system? Let’s start with what’s clear: the current approach isn’t working, especially at the local level, for reasons outlined below. Yet stores, and the supply chain as a whole, are now more connected than ever. These interconnected data points and touchpoints create new possibilities for the store to serve at the speed of today’s hyper-connected shoppers, allowing retailers to delight their customers and grow their business.

The current planning process is outdated and broken

Most retail planners recognize traditional top-down planning diagrams. They start with a big-picture view of the planning process, which can be broken down by geography or brand, grouped into store clusters and then spread like peanut butter down to the local store. Layer in some promotions to drive certain categories and products. Run sales and specials for things that don’t move. Worst case, make it someone else’s problem by sending the goods back to the DC, try to get a credit from the vendor or send the unsold goods to another store in hopes that they will sell better there. Repeat the process on an annual, quarterly or monthly basis, and what you end up with is a wasteful glut of unwanted inventory taking up precious space in the store.

It didn’t take long for digital technologies and e-commerce to show that there was a better way, with offerings like buy online and pick-up in store (BOPIS, otherwise known as “click and collect”). Retailers could offer nearly unlimited options to an online consumer and have them ready to be picked up in the store, or else deliver to wherever they happen to be. What if we could adopt this same flexibility to create the best inventory assortment in the store?

Serve at the speed of the customer

With AI and data, stores get a whole lot smarter

It wasn’t long ago that store profitability was driven by the speed with which shelves could be stocked and people checked out at the register. Stack it high and let it fly, as all industry veterans have heard. But when what we stacked high was no longer the right product, customers started to look elsewhere. They jumped on their smartphones, trying to check stores for the items they wanted before they showed up in person. Eventually they just gave up and started buying online. This led to major over-ranging of product in stores because retailers were trying to be everything to everyone.

An entire ecosystem of software companies has grown up around treating the symptoms, rather than fixing the root problem. Markdown optimization, for example, emerged because we were all working with rear-view data, and a one-size-fits-all static solution. But with new streams of data and improvements in technological capabilities, such as deep machine learning (ML), operations research (OR), time series and vision (OCR)⁠—as well as with explainability, which allows AI to learn from interactions both with data and people⁠—a new AI-powered approach is possible.

AI allows retailers to drive SKU-level profitability locally, as if you had your best planner at each store. Using whatever data the store has, whether it’s point of sale (POS), inventory or demand forecasting, retailers can build much more accurate plans. It will also be possible to layer in edge data from IoT, competitive information, local events, weather and cameras, as these different streams of data open up and start to flow.

AI can make sophisticated recommendations, and it learns and gets smarter with every interaction and new piece of information it processes. The speed of the entire supply chain can then be leveraged⁠, whether it’s neighboring stores, an e-com service point or a regional DC to solve the last mile and potential disruptions. The juice is worth the squeeze, as they say, because an interconnected engine that makes the right prediction and recommendation, and can even take action within a specified time horizon, allows for the kind of real-time orchestration that will give customers what they want, when and where they want it. This is how stores are becoming the engine that drives agility in the retail planning process.

Use stores to serve at the speed of the customer

Today, it’s the customers who run the show, and the retailers who obsess over them win the game. The most successful retailer of our time (yes, retailer) predicted this about 20 years ago. The challenge up until now has been that stores have struggled to keep pace with the evolving consumer and technology landscapes. The decreasing cost of in-store tech, combined with the ability to make sense of the truly massive amount of data now available to retailers, makes it possible to move stores back into the lead position as a brand’s most strategic asset. This is made crystal clear by the fact that brands that started as online-only are expanding to physical stores—just look at Amazon Go, the checkout-free stores the company is currently rolling out across the U.S.

Stores also play an important role in brand loyalty: one recent report found that 77% of consumers say they’ve had relationships with specific brands for more than 10 years, and in a vast number of cases, these relationships started in the store. AI makes the store even more valuable by harnessing the power of traditional and edge data and leveraging new insights, decisions and actions to drive space and product profitability by store and by SKU, every day. With customers spending more per visit in stores than online, according to one survey, there’s a big incentive for retailers to get this right by leveraging data, AI and the interconnected supply chain.

Retailers that embrace a digital, interconnected supply chain that will emerge as the world’s top brands, because they can serve customers at the speed necessary to compete with online-only options. This interconnected world starts with the store, but goes much further. The more sophisticated a retailer becomes at connecting the touchpoints, the more the store, the space and the assortment benefit, at every location.

This can seem like a lot to digest, and one may ask, as the saying goes, how do I eat this elephant? The answer is very clearly one bite at a time—starting with the store, where OOS detection is the best starting point to providing a better inventory assortment at the local level.

Set off on your AI journey, one store at a time

Leveraging data, AI and the whole interconnected supply chain may seem like a daunting task for many retailers, but there is a proven framework for success. A phased approach allows for value to be unlocked with each store. It begins with detecting inventory patterns. This visibility is the right starting point, but unlike traditional control towers, an AI-powered solution goes further, predicting, prescribing and ultimately executing a better assortment.

Journey to AI supply chain success.

Using available data to predict OOS and inventory flow allows us to prescribe the right products and range at each store, and build and execute a new assortment plan that makes the local store the centerpiece of the process. These localized insights are used to increase the accuracy of the plan that’s created at, say, a cluster, category, brand or national level. This ensures the right SKU is on the right shelf at the right place and time, at all times, to drive profitability.

Until recently, there was no way for retailers to capitalize on the data available at the store, and our supply chains were neither nimble nor fast enough to be able to respond at the speed of the customer. Now, with the advent of new technologies and active learning systems, there is a bold opportunity for those who are ready to seize—and it all begins with the store.