Now and then, I meet a unicorn — that rare person who truly doesn’t enjoy a stroll through a “real” store in search of an adventure — culinary, sporting, fashion, gardening, literary, musical…. The sense of discovery and surprise at something that matches an unrecognized yearning can be exhilarating. But the sheer size of many stores is exhausting — with some, I need a site map, flashlight, and overnight bag to take it all in. In others, the inventory is specialized and packed in so tightly that online seems a better way to find what I want.
If you’re primarily an online shopper, you may be asking how relevant brick-and-mortar stores are. The answer is: quite relevant. According to the latest U.S. Department of Commerce report, 88 percent of U.S. retail transactions (still) occur in traditional brick-and-mortar stores, where customers, employees, and inventory come together.
But the industry is morphing. Old brick-and-mortar brands that we loved are consolidating with other brands or simply disappearing, primarily due to demographic shifts and the success of “click and ship” experiences. Those retailers who are transforming in recognition of these shifts often share common challenges that fundamentally center on how to increase revenue and profitability. Retailers need to deliver a better customer experience and build loyalty to take a larger share of customers’ wallets, and they need to squeeze the excess costs out of every process, from marketing to supply chain to checkout and delivery.
With these goals in mind, retailers are transitioning to a customer-centric environment of “everywhere commerce,” with a blurring of the borders between in-store and online shopping. Consumers want a personalized, seamless shopping experience that includes high-touch, hyper-personalized customer service across all sales points. Retailers want to deliver a safe and enjoyable shopping experience for both their employees and customers. Meeting these objectives requires providing a connected experience at the stores with context, and in real-time, from the moment customers arrive at the curb.
The traditional way of delivering superior customer experience is human-intensive — relying on traditional systems and processes that are skills-driven. Improving the experience typically drives costs up while limiting scale. Retailers need to deliver solutions with a higher level of accuracy — that meet more customers’ needs more often — but at lower cost, and with the ability to rapidly scale to hundreds of stores to enhance the brand.
How AI can help
Retailers are increasingly turning to AI-driven applications that enable a broad range of solutions to enrich the shopping experience and predict what individual customers want, and when and where they will want it. AI is also helping retailers automate time-consuming processes, enable smarter inventory management, maintain defenses against unintentional and fraudulent causes of inventory loss, and deliver a safer shopping experience.
AI solutions can run in almost any location; however, given the focus on sensing, processing, and responding instantaneously to situations, events, and actions at the store (the “edge”), edge-based technologies are well-suited for many retail use cases. Today’s solutions can operate securely and at scale and be deployed, managed, and controlled remotely.
To learn more about how retailers are using AI at the edge, I caught up with Chandra Venkatapathy, Director of IoT Solutions at Dell Technologies, who works with retail customers worldwide to grow their businesses with the help of edge solutions. He offers the following examples of specific ways in which AI-driven applications help retailers address retailers’ challenges.
With the power of analytics and AI, modern retailers can leverage data from many channels to understand everything from customer buying behavior to product trends and product pricing optimization. AI can also help retailers gain a clear view of how much product to stock and which products to suggest to particular customers or demographic groups.
“The key is using intelligent systems to understand customers at a personal level — who they are, when they arrive, where they go, with whom they interact — to anticipate their needs and suggest the right products at the right time,” Chandra notes.
Empty store shelves and out-of-stock products are costly in terms of lost revenue and disappointed customers. Overstocked shelves can also be expensive. AI systems can help retailers avoid these problems by monitoring inventory levels, predicting purchasing patterns, and automatically ordering goods to replenish shelves. With better predictive capabilities, retailers can keep inventory at an optimal level.
Retailers face loss risk from many sources, but one area where many are using AI to combat shrinkage is at the point of checkout. A survey by the National Retail Foundation found that more than 55 percent of the retailers who responded are implementing point-of-sale analytics to combat problems such as label switching and scanning errors. Some are also implementing video analytics, fingerprint ID, and facial recognition solutions to stop repeat offenders.1
For today’s retailers, in-store safety of employees and customers is a high priority. AI solutions that employ computer vision and other advanced technologies address emerging needs. For example, solutions can check for thermal variations at the entrance for a safer shopping experience, verify social distance compliance within the store, and count people for fire code compliance.
Primary considerations for capitalizing on AI
In our conversations, Chandra offered suggestions for retailers wanting to capitalize on AI.
Focus on the problem and the process.
“In developing AI applications, you must start with the desired outcome — the result you’re seeking — more so than the technology you will use,” Chandra advises. “Focus on how your processes will be optimized and improved.”
He suggests starting with an “as-is” process and working to identify the areas of how and where to improve to get a clear view of the “to-be” state. That outcome will help a data scientist choose the optimal AI model, along with associated parameters, such as inference metrics for accuracy, scale, and latency needs, which in turn drive the technology infrastructure requirements.
Choose the right use cases.
AI-enabled solutions are ideal for repetitive use cases that require processing voluminous data at a short time, or that require hyper scale, Chandra notes.
“Though humans can solve many of these use cases, manual processes tend to be expensive, error-prone, and slow,” he says. “With recent developments in computer vision, AI enables retailers to deliver solutions at higher accuracy and lower cost than ever before.”
Proceed in a stepwise manner.
Chandra advises against leaping directly into large-scale AI development projects. Instead, he suggests starting with a clear plan focusing on technology validation, process validation with employees and existing systems, the ability to deploy at scale, and security.
“Many clients start at the lab for proof-of-concept (POC) and then graduate to a store environment for proof-of-value,” he says. “Based on those results, clients extend these projects to a multi-store POC that is a reflective sample of various regions and store types. Adapting to lessons learned from that POC enables successful expansion to large-scale deployments.”
Get a jump start on your project.
When putting the systems in place for AI, the question of “buy vs. build” is a challenge for many organizations. Chandra suggests thinking in terms of time-to-value and how you can best keep up with the latest innovation while aligning with your strategic interests.
“Many organizations find that they can get AI-driven applications up and running faster with solutions that bring together server, storage, and networking components that are pre-tested and validated with the software for the planned use case,” he says. “These new solutions can significantly simplify technology planning and deployment while accelerating time to value.
Surround yourself with the right ecosystem.
Finally, keep in mind that AI projects are team efforts that require a lot of partnering with people who can help you put the pieces of the AI puzzle together.
“Don’t try to go it alone,” Chandra says. “Work with technology partners who have the engineering expertise and hands-on experience to help you put together solutions that can dramatically accelerate your success with AI, machine, and deep learning environments.”
The retail industry is in the midst of a sweeping transformation driven by new customer expectations, new technologies, new business models, and an ever-changing environment. Artificial intelligence is a critical enabler to help retailers transform successfully.
For retailers who are poised to seize the day, AI-driven systems offer a clear path to more personalized customer relationships, higher sales volumes, automated inventory processes, and smarter approaches to loss prevention. These outcomes require organizations to recognize the opportunities, focus on the results, and put the right technologies in place to move forward.
To learn more
For a deeper dive, see the Dell Technologies solution brief “Revolutionizing retail with integrated point of sale (POS) solutions.”
1 National Retail Federation,“2019 national retail security survey.”