Photo: Walmart
Walmart puts AI to the test in an in-store lab
Walmart is opening a laboratory inside a small Walmart in Levittown, NY to test artificial intelligence (AI) applications for both associates and shoppers.
The Intelligent Retail Lab (IRL), according to TechCrunch, will see how AI can be used to identify low stocks on shelves, when items are on the wrong shelf and spillages. Walmart is also looking for a better understanding of when shopping carts are running low near the store’s entrance, according to the report.
Hardware, software and other equipment have been installed, but the lab is not yet operational. IRL is part of Walmart’s incubator Store No. 8. and being led by the Kepler Project. The Kepler team, according to past reports, is also testing computer vision and cashier-less technologies similar to the Amazon Go experience.
In late October, Sam’s Club said it was opening a cashier-less concept in Dallas that enables shoppers to use its Now app to not only pay for purchases, but access smart shopping lists, store maps and augmented reality tech to access product information. Like IRL, some features support store operations.
“We’ll test electronic shelf labels that will instantly update prices, removing the need to print and replace new item price signs,” said Jamie Iannone, CEO, SamsClub.com and EVP of Membership and Technology, in a blog post. “And down the road, we’ll use the more than 700 cameras in the club to help us manage inventory in new ways and optimize the layout to make shopping effortless.”
Many emerging AI-powered solutions promise to bring automation and advanced forecasting to in-store functions while working in sync with robotics, RFID and other technologies.
Afresh Technologies, a start-up based in San Francisco, for instance, uses machine learning and AI to improve demand forecasting to help optimize in-store replenishment of fresh food. Founder Matt Schwartz recently told The Spoon, “We believe there’s a dearth of intelligence in the fresh food supply chain, and as of now inventory solutions are often really inaccurate.”
-
Walmart Opening AI-Focused Retail Lab In A Long Island Store – TechCrunch
-
Walmart Opening AI-Focused Retail Lab In A Long Island Store – Retail TouchPoints
-
Sam’s Club Now – Reimagining the Future of Retail – Sam’s Club
-
Is Sam’s reimagining the future of warehouse clubs? – RetailWire
-
Afresh Uses AI to Optimize Fresh Food Stocking and Cut Grocery Waste – The Spoon
-
Inventory Management Now a Key Battleground in the Digital Transformation of Brick and Mortar Retail – ABI Research/PRNewswire
-
New Artificial Intelligence-Enabled Solution Helps Retailers Improve Demand Forecasting Accuracy, Increase Sales and Reduce Waste – Symphony RetailAI/Globe Newswire
Discussion Questions
DISCUSSION QUESTIONS: Will artificial intelligence be a game changer for inventory optimization, labor scheduling or other aspects of in-store operations? Will the biggest such AI-benefits come from enabling automation to reduce repetitive tasks or in areas such as forecasting and problem solving?
Poll
BrainTrust
Carol Spieckerman
President, Spieckerman Retail
Sterling Hawkins
Co-founder, CART
The ability to automate shelf replenishment has been around since the late 1980s – we were doing it at the Retail Alliance. All it required was a starting inventory, an effective POS system, and conscientious entries when shelves were re-stocked. There’s nothing AI in this, so no – it will not be a game changer. Remember – AI is not automation.
Artificial intelligence is the most promising innovation in retail as it has so many applications. Ranking the benefits misses the point since all of the tests Walmart is running will ideally work synergistically. Through dedicated laboratories like the Sam’s Club Now location in Dallas, Walmart will be able to test the full potential of AI on multiple fronts then deconstruct and deploy what works rather than having to tease out the insights from its regular formats. Things are going to start to get exciting on the AI front, just in time for the holiday season.
AI will not be a game changer for inventory management. There is nothing artificial about inventory management within the four walls of the store. Centralized inventory management, replenishment, and demand forecasting systems are able to do the job fairly accurately, provided there is the commensurate amount of in-store discipline. There may a play for AI in labor scheduling however.
I am with Bob and several others on this one. Not seeing AI as an inventory solution. For my 2 cents.
The applications mentioned all seem to lean towards automation, not AI machine learning – the majority of those kind of opportunities would seem to be on the forecasting/problem solving/opportunity discovery side.
I’m not sure if I buy Walmart’s motives. Surely there is a less expensive fix to solving the problem of “understanding when shopping carts are running low near the store’s entrance”?
Still, I am all for brands who are actively searching for efficiencies and new uses for technology, so we will see what Walmart comes up with!
Out-of-stocks remains a big problem for many retailers and it is a problem worth looking at with AI. For Walmart in particular, I think they once calculated the cost of this problem at about $3 billion a few years ago. The challenge is in the activation of the AI — the operationalization of the analytics to automate processes. This has proven the biggest barrier to many use cases. For the use case underway at Walmart there are a lot of moving parts, human and technical. Companies taking a serious approach to AI but lacking the resources and will of Walmart might think smaller and more focused on improved human decision making with AI. Later with greater resources and expertise, it will make sense to automate related business processes.
AI is up and coming and will become more prominent as companies like Walmart experiment and validate investment returns. The key places where AI has seen success in retail is in demand planning and inventory management. Other companies are trying to find ways to map in-store labor, but these solutions are more business-rule focused than mapping to predictive AI systems. The challenge remains the dynamic nature of such solutions. Forecasting will drive AI value, but will take time to expand to specific problems first, then later expand more generically to issues. There will be some applications in personalization as well, especially as part of larger commerce solutions. However, it will impact the back office and operational side long before consumers know it’s there. Conversational commerce might be another public avenue that has potential — but even here the returns are not immediate.
Machine learning, artificial intelligence and augmented reality are quickly becoming the new omnichannel buzz words for 2018 and, of course, will be prominently featured at the NRF in January. Automation has its place and it will certainly permeate around retail organizations in the next five to 10 years. Walmart along with Amazon are two of the more leading innovation-first companies that are truly experimenting with AI. Taking a crawl-walk-run approach is the way to go before scaling this out.
However where the rubber hits the road is, how exactly do retailers and CPG companies actually operationalize their strategies with all the richness that AI and all these innovative technologies offer? With any innovation and change of this magnitude, there has to be solid executive sponsorship to help drive the culture of change in these organizations.
AI has so much promise and potential, however, we should temper our enthusiasm until organizations mature and change to really make an impactful difference in this space.
I am very surprised that Walmart looking to test AI to improve inventory and on shelf availability is making the news. AI or more accurately machine learning has been a part of state of the art inventory management and merchandising systems for a while now; it is only that the consultants have jumped on AI as their next big payday that it has become the topic of every retail conversation. Machine learning is key in improving forecasting for promotions, weather related items and, particularly in fresh and short life products, improving availability whilst reducing shrink and waste.
Yes, there are constant developments being made and the rapid improvement in RFID accuracy is helping to drive further improvements in the supply chain, the more accurate the data the better the analytics leading to better results. The advent of robotics will be another possible step forward and there are some exciting new developments in this area, some good and some not so good. But to claim that Walmart is of ahead of the game in using AI for inventory management and on shelf availability is missing so much great stuff that is already out there and available to all retailers. They just need to invest in modern state of the art solutions and not sit on their old legacy solutions.
AI is not the game changer for retail and inventory management. Empowering consumers, and using all of the data available to the retailer (in even the most basic applications) are the keys to moving retail to the next level. To do this, retailers do not need AI, but need to apply simple automation, and trust. Staying abreast of store inventories, self-checkouts, and reaching out to the consumer are the keys to success here. Focusing on applying simple inventory and supply solutions to eliminate out of stocks, merchandise shelves correctly (which AI does not do), and price products appropriately (another non-AI function) are important.
Store-level inventory optimization is the “cutting-edge” innovation that has been honed in grocery for close to 20 years. After some early attempts to automate ordering using computer assistance proved disastrous, the grocery industry mostly turned its back on CGO/CAO solutions.
A handful of chains in North America saw it differently. Price Chopper (Golub) in New York State was an early adopter that implemented its solution about 17 years ago and has enjoyed ongoing benefits of minimal out-of-stocks, minimal excess inventory, and more efficient store delivery. Today, Sobey’s and Wegmans are also among the success stories.
One element described here that none of these retailers has needed to get its forecasts right is the 700 cameras that Mr. Iannone has planned for a Sam’s Club. There are other practical ways to keep accurate tabs on inventory levels as part of the routine receiving-merchandising-stocking process.
In-store sensing is absolutely a revolutionary area in retailing. Store-level forecasting and automated reordering pays great dividends in replenishment categories. Machine learning is a requirement to make these things happen, but promoting AI as the panacea is as much hype as substance.
AI is going to play a vital role in the future of CPG. At the end of the day, the biggest impact will come in improving the customer experience. Let’s face it: Consumers are demanding and they have a right to be so. They are laser-focused on accomplishing their mission — buying the product they want, where and when they want it. By embracing Big Data, machine learning and artificial intelligence, retailers can already quickly and easily identify critical issues with in-store execution that prevent the product from being on the shelf when the shopper comes to make a purchase. Simply alerting in-store staff of some of these simple errors — such as misplacement within the plan-o-gram, product that is blocked by another item on the shelf and/or missing price tags — allows for quick adjustments that will get the right product to the right place at the right time, providing an excellent shopper experience and, ultimately, strengthening loyalty.
A retailer with 50,000 SKUs requires 6×10^3600 calculations every week just to maximize revenue in the selection of weekly circular items. To put that in perspective, the visible universe is estimated to contain between 10^78 and 10^80 atoms. AI is capable of processing at scales that humans find hard to understand and are incapable of doing alone. The lowest hanging fruit (where it’s already starting) is in very focused areas with defined problems/optimizations such as pricing and inventory management. Over time, AI is a game changer for all of the business areas above and more. It will literally transform business itself.
Whether what Walmart is described as doing really counts as AI aside, I like the idea of testing new ideas via an in-store lab. There’s really no better way to see what works and what doesn’t than putting it in a live environment and if Walmart can do this via the lab it will learn plenty. Knowledge is power after all and although the applications sound like obvious choices if Walmart can gain a better understanding of its spaces via tech then maybe they can effectively solve some of these issues.
AI can be a helpful advisor to many aspects of store optimization, and it is good that Walmart is experimenting and learning about AI for store operations. However, AI is not perfect, so there will still be errors made in the operations, but less so. Forecasting and problem solving will be better beneficiaries of AI than reducing repetitive tasks. Forecasting and problem solving will lead to better store operations with one of the benefits being a reduction in repetitive tasks.