In the decade following the Great Recession, a “retail apocalypse” decimated the retail industry, resulting in the closing of thousands of brick and mortar stores.
While the past 10 years have been disastrous for physical stores, the pace at which stores have closed has continued to accelerate, reaching record levels. Retailers closed over 9,300 stores in 2019, according to an analysis by Business Insider, breaking the previous record of 8,000 store closures in 2017. Cushman & Wakefield forecasts as many as 12,000 stores could close in the U.S. this year, easily smashing 2019’s record.
To combat this carnage, struggling retailers have been turning to technology to lure shoppers into their brick and mortar stores, and keep them spending in-store instead of online. The technologies enabling the retail store of tomorrow will create a completely different shopping experience than today, but much of this nascent digital transformation is focused on the basics, such as inventory management.
Getting the basics right matters, because just one bad experience can hurt a retailer’s bottom line. ServiceChannel’s “The Brick and Mortar Retail Report” finds that physical stores are often un-shoppable, with two out of five shoppers saying they were met with empty shelves or had trouble with disorganized inventory. The survey reports that after a negative experience, 40% of shoppers spend less money in a store, 52% of them will leave without making a purchase, and 69% are less likely to return.
Autonomous Perception Systems for Retail Inventory
One company providing retailers with breakthrough technology for better visibility into managing their inventory is Pensa Systems. Keeping the right products on the shelf is an age-old problem, according to CEO Richard Schwartz, and stockouts (when items are out of stock) are a $1-trillion problem for retailers and the brands who sell through them.
“One in eight products is not on the shelf, available for purchase,” Schwartz says. “The problem is solved today by walking around and staring at the shelf and seeing what isn’t there.”
Any ability to automate a solution for this issue is a game-changer for the industry. It turns out many technologies have been tried: cameras in the ceiling, fixed cameras, ground-based robots, and more. These solutions are often expensive, disruptive for the store, and sometimes not even accurate, Schwartz points out.
Pensa’s novel automated solution for inventory management uses computer vision, through autonomous drones and mobile cameras, to understand what products on store shelves, down to the individual item level, and artificial intelligence and neural networks on the back-end to manage inventory. “It is a combination of a new technology, a new way to deliver it, and a painful problem with an immediate business value; we just put it together,” says Schwartz.
The drones do not transmit any data while flying, waiting until they land on their recharging station, which is also an edge-computing device. Almost no data processing is done on the drone; some is done on the edge device, but most of the data is processed on large-scale distributed systems in the cloud.
“In a very accurate way, we can report what is out of stock and what is running low,” Schwartz says, adding that Pensa’s 98+% accuracy rate for out-of-stock detection aids retailers in optimizing their inventory for a better shopping experience.
Seamless Shopping Experiences
According to a recent survey by retail consultant BRP, 96% of shoppers say ease of checkout and payment are important factors when choosing where to shop. To that end, AWM Smart Shelf created a seamless shopping experience. Using artificial intelligence, computer vision, and machine learning, the company has developed a cashier-less shopping model called AWM Frictionless. The platform detects which products shoppers are removing from shelves, and adds them to their shopping cart for easy walk-out shopping, with automatic billing to their digital wallets.
“We developed some in-house sensors that are computer vision-based that allow us to accurately track what customers are picking up, putting down, and taking with them,” says Kevin Howard, founder and CEO of AWM Smart Shelf. “It is a walk-in, walk-out solution that ultimately will bill a shopping cart.”
Howard explains that within the shopping cart, the company is starting to implement AWM Facial Wallet, a face-pay application in which shoppers can use their faces to access the store. “Your face is tied to a digital wallet, so you don’t need a digital app,” he says. “You can just come and go in a seamless fashion.”
The expectation is that AWM Smart Shelf eventually will build a network of retailers that will utilize this facial wallet application. According to Howard, there is going to be a significant shift around food and fast-moving goods into fully autonomous micro-markets, which are unmanned stores that have small, localized footprints similar to AmazonGo, Amazon’s cashier-less stores. Consumers will be able to seamlessly shop in department stores, grocery stores, pharmacies, big-box retailers – basically, anywhere consumers shop today.
Ultimately, Howard thinks store openings will outpace store closings, but they will be in small-format footprints, i.e., these micro-markets.
A Glimpse into the Future of Retail
Walmart’s Intelligent Retail Lab (IRL) is an artificial intelligence (AI) laboratory located within a 50,000-sq.-ft. Walmart Neighborhood Market in Levittown, NY, and is focused on uncovering applications of AI within physical retail.
“While the application of AI in e-commerce is now ubiquitous, there haven’t been many real explorations for the potential of these technologies in a real-world retail environment,” says IRL CEO Mike Hanrahan. “IRL exists for exactly that reason, bridging the world between the digital and physical realms to enhance the Walmart associate experience and simplify the shopping experience for customers.”
Leveraging both hardware and software, Walmart is testing technologies including AI, computer vision, machine learning, and sensor fusion. To fully capture these efforts for study and improvement, the store is equipped with a range of camera hardware and product intelligence technologies that are connected by enough cabling to scale Mt. Everest five times, and enough processing power to download three years’ worth of music (27,000 hours) each second. “Our in-store data center operates with 100 servers, 400+ graphic processing units, nine cooling towers, and 12 intermediate distribution frames for networking,” Hanrahan adds.
The initial area of focus at IRL is improving inventory and on-shelf availability. A combination of cameras and real-time analytics automatically triggers out-of-stock notifications to internal apps to help associates determine when items need to be restocked and organized.
For Hanrahan, IRL’s tests of emerging technologies will identify capabilities that can contribute to the evolvution of the retail experience for both associates and consumers.
“We operate like an AI factory where we test and learn, and then scale specific AI products to additional stores once they’re deemed successful,” Hanrahan says. Over time, the goal is to leverage this continuous experimentation to scale enhancements to the customer and associate experience in Walmart stores across the country.
John Delaney is a freelance writer living in New York City, NY, USA.