Eye on AI - July 2nd, 2021
Welcome to Aigora's "Eye on AI" series, where we round up exciting news at the intersection of consumer science and artificial intelligence!
This week, we’ll be looking at Walmart’s development of a grocery substitution recommendation AI meant to bolster e-commerce sales, then shift to break down Amazon’s holistic (and replicable) “flywheel” approach to AI usage.
Walmart Enlists AI for Online Grocery Substitutions
We begin with the news that further highlights the continuing trend of customers transitioning to online shopping. According to a recent Supermarket News article, “Walmart enlists artificial intelligence for online grocery substitutions,” the retail behemoth has announced the creation of an online grocery shopping recommendation AI that gives alternative product recommendations to customers searching for out of stock food items. The move is meant to bolster order accuracy and customer satisfaction for e-commerce grocery sales that have continuously risen since the start of the pandemic.
“Instead of having to guess, the personal shopper can be told precisely what the customer may prefer,” says Srini Venkatesan, executive vice president at Walmart Global Tech. “If our personal shoppers are preparing orders and come across an item that is not available, our system suggests the alternative product. Our tech even shows our personal shopper where the item is located in the store, simplifying the decision-making process for our team and enabling them to prepare orders quickly and efficiently.”
The solution recommends the next best product in stock-specific to each individual customer. Preference is personal. Taste and brand preferences are as unique as the human race which makes it impossible to denote any one alternative product as the “best.”
Factoring this idea into its deep learning AI, Walmart’s scientists trained its recommendation system to consider hundreds of variables in real-time, including size, type, brand, price, aggregate shopper data, individual preference, and current inventory, etc. to determine the best alternative item for any given online customer. Once a recommendation is made, customers are asked to approve or reject the substituted item with each decision being fed back into learning algorithms to improve the accuracy of future recommendations.
To say that results have been encouraging is a bit of an understatement. Since deployment, Walmart’s customer acceptance of online grocery substitutions climbed over 95%, while more than fifty percent of consumers shopping on Walmart.com purchased groceries over the first quarter of this year compared with 23% for Amazon. That makes Walmart’s online grocery shopping presence even stronger than Amazon’s! With Amazon mostly dedicating its grocery AI efforts to in-person shopping (see cashierless Fresh grocery stores), its online grocery presence has suffered. Since Q4 of 2020, e-commerce grocery purchases have dropped nearly eight percentage points. Still––based on what we know about Amazon’s comprehensive AI approach, I doubt Amazon is worried too much.
The Amazon “Flywheel” Approach to AI
Speaking of Amazon’s AI approach, do you remember the flywheel, the wheel on a potter or sower’s machine that gains momentum each time the foot pedal is pushed? The basic idea is this: the faster the wheel goes and the momentum it gains, the easier it becomes to press the pedal and add even more speed.
In engineering terms, the “flywheel” approach describes a way companies can conserve energy as they keep up the momentum. And, according to Bernard Marr’s article “How Amazon Uses Artificial Intelligence: The Flywheel Approach,” no one uses this approach more brilliantly than Amazon, especially when it comes to implementation of their AI.
“Amazon’s entire organization is constantly humming with artificial intelligence, and founder Jeff Bezos mandated that data is shared across the organization, not hoarded in one department or process,” writes Marr. “Datasets are always connected to other data in the organization, to make sure they can be externalized from the ground up. The flywheel of continuous data and AI keeps different parts of the Amazon engine going, and innovations that take place in one department or team can be transferred to other parts of the organization.”
Consider all the ways Amazon is utilizing AI and collecting data: There’s conversational AI in Alexa, which improves speech tech and collects data from shoppers while they’re at home or on the move; there’s the e-commerce recommendation engines, which drive 35% of Amazon’s total sales. There’s search data and inquiry data, in-store data from Amazon Fresh and Go and e-commerce data. There are automated smart robots stocking Amazon’s warehouses with goods and smart vehicles and drones being developed to deliver them. And all of these systems are sharing data with one another to improve.
No retailer collects the amount of data Amazon does. And no retail company, at least at the moment, can threaten Amazon’s dominance of retail AI. But that flywheel approach can (and should) be replicated. And the sooner that process is started, the more momentum that flywheel builds and the easier it becomes to add momentum with just a little bit of effort.
That's it for now. If you'd like to receive email updates from Aigora, including weekly video recaps of our blog activity, click on the button below to join our email list. Thanks for stopping by!