Skip navigation
With Machine Learning, Amazon and Whole Foods Could Satisfy Every Customer Craving Joe Raedle/Getty Images

With Machine Learning, Amazon and Whole Foods Could Satisfy Every Customer Craving

The Amazon and Whole Foods deal could pave the way for broad adoption of machine learning in grocery retail. 

Many Whole Foodies were left feeling like they drank a bad batch of kombucha last week at the news that retail behemoth Amazon would acquire the healthy grocery chain for $13.7 billion. The deal, if it goes through, will certainly have an impact on the retail industry (one that has already been felt on the market), but its impact on the tech industry – in particular, the use of machine learning in supply chain management and more – could be profound.  

Whole Foods began its journey to the cloud around 2013, with a plan to replace 80 to 90 percent of its systems by 2020, grouping it into three major platforms: its team member platform, its customer platform, and its product platform. Of course, this plan could be fast tracked now that it has the world’s biggest cloud provider as its parent company. Amazon may have extra motivation to move quickly as Whole Foods currently uses a number of solutions from one of its biggest cloud rivals, Microsoft.

Speaking at the Oppenheimer 16th Annual Conference last June, Whole Foods CIO Jason Buechel said that the company is “leveraging new advancements in machine learning, Big Data and we’re doing it in an architecture that allows us to put all this in the cloud…Gets us away from batch jobs, things that are very much 1990s architecture.”

“So I think the more exciting piece is the ability to automate stuff that we do today and ultimately actually use data and analytics to better support our business,” Buechel said at the time.

Amazon CEO Jeff Bezos said that much of what Amazon does with machine learning “happens beyond the surface.”

“Machine learning drives our algorithms for demand forecasting, product search ranking, product and deals recommendations, merchandising placements, fraud detection, translations, and much more. Though less visible, much of the impact of machine learning will be of this type -- quietly but meaningfully improving core operations,” he wrote in a letter to shareholders in April.

But Amazon Go is perhaps its most ambitious use of machine learning yet, at least in how it applies to the Whole Foods deal. Amazon Go is a grocery store that has no cashiers, and all shoppers need is the Amazon Go app to start shopping. Its system uses computer vision, sensor fusion, deep learning, and sensors to determine which products customers are putting in their bags, charging the customer automatically through the app, in what it calls Just Walk Out Shopping.

On its website, Amazon says, “Four years ago we asked ourselves: what if we could create a shopping experience with no lines and no checkout? Could we push the boundaries of computer vision and machine learning to create a store where customers could simply take what they want and go? Our answer to those questions is Amazon Go and Just Walk Out Shopping.”

Machine learning in the fight against food waste

Machine learning is already helping grocery stores minimize food waste and increase in-store product availability, according to McKinsey.

 “By using predictive applications powered by machine learning, an international supermarket chain with more than 1,000 stores automated most of the central planning and decision making for daily orders in one of its largest fresh-food departments,” McKinsey said in a November report. “And because the retailer operates several food-processing plants, it was also able to integrate warehouse and manufacturing processes—for instance, through just-in-time production—to reduce stock in the entire supply chain, increase in-store product availability, and get fresher products on store shelves.”

In a study by Blue Yonder, a developer of cloud-based predictive applications for retail, found that replenishment is one of the biggest challenges in terms of fresh grocery, and 46 percent of grocery directors said that they left their replenishment decisions to gut feeling. The study found that retailers using machine learning have seen a reduction of up to 80 percent in out-of-stock rates without increasing waste or inventory.

Back in 2015, Whole Foods started using the Infor CloudSuite Retail platform for a variety of strategic merchandising and supply chain management functions. The technology platform uses predictive tools for retailers “to get ahead of consumer demands.” Last year, Infor acquired Predictix to continue to build out its suite of enterprise applications that intersect cloud, analytics, machine learning and self-service, Predictix CEO Molham Aref said.

"From my point of view, long in the tooth time-series approaches to forecasting and analytics that inform pricing, assortment, promotion, and other demand shaping tactics have run their course,” Greg Girard, program director of omni-channel retail analytics at IDC Retail Insights  said in a statement last year. “They fall far short of what's needed today. They were designed and tuned for a different era-before elastic, scalable, subscription cloud environments, operationalized machine learning, inexpensive data storage, and the complexities, scale, and speed of omni-channel retail. Going forward leading retailers will take advantage of these advances for financial benefits, operational efficiencies, and customer loyalty."

Machine learning use-cases in grocery

Last month, grocery delivery company Instacart – who happens to be both a partner of Whole Foods and Amazon – released its first public dataset, an anonymized dataset that contained a sample of over 3 million grocery orders from more than 200,000 users. It said that it hopes that the “machine learning community will use this data to test models for predicting products that a user will buy again, try for the first time or add to cart next during a session.” Instacart uses XGBoost, word2vec and Annoy in production to sort items for users to buy again and recommend items for users while they shop.

In the U.K., online-only grocer Ocado – which saw an 11 percent bump in its stock this week on the Amazon/Whole Foods news – uses machine learning to monitor demand for products and use the information to map out an optimal storage scheme, according to a report by MIT Technology Review. In addition, machine learning is used to “spot missing items in a shop, populate a basket of groceries on the basis of learned preferences, and even suggest versions of products that are lower in salt or sugar.” At least one of the company’s investors sees an opportunity in rebranding as a tech company and letting other grocery retailers use its smart software, but for now its focus is on its own retail business.

Beyond its usage in ensuring produce is fresh and learning customer preferences, machine learning could also help prevent online outages through anomaly detection. Grocery downtime can spell disaster for customers, just ask U.K. grocery retailer Tesco, who had to cancel thousands of orders this week after a fault with software used by its workers who pick customer orders from stores. 

While much of the focus on machine learning is its role in replacing humans, using it properly could create an amazing experience for customers – which fits with both Amazon’s and Whole Foods’ approach.

“There are many advantages to a customer-centric approach, but here's the big one: customers are always beautifully, wonderfully dissatisfied, even when they report being happy and business is great," Bezos wrote in April. "Even when they don't yet know it, customers want something better, and your desire to delight customers will drive you to invent on their behalf. No customer ever asked Amazon to create the Prime membership program, but it sure turns out they wanted it, and I could give you many such examples.”

If customers won’t tell you what they really want, machine learning may be the only way to truly satisfy them. 

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.