Walmart Stocks Up on AI-Fueled Insights
From the cloud to the C-suite, AI is (quietly) reinventing business decision making across industries. One buyer? The world's largest retailer.
News articles, millions of them. Research studies. White papers. Every blog, tweet and social media update. Imagine you’re a CEO charting your company’s short- and long-term business plans. You've got access to every scrap of content published since 2013, and you’re eager to scour it for emerging industry trends.
Got a year—or a century—to spare?
Of course not. That’s where AI-focused market research firm Quid comes in. When Bob Goodson left his sales manager position at Yelp in 2007 (fun fact: he was the company’s first employee), he founded Quid with a vision to build an AI platform that could analyze and visualize the world’s written content in a way—and at a volume—that humans simply couldn’t.
“The goal was to create a better way of consuming content with machines,” Goodson says. “Quid’s report is designed to be akin to something that you would create if you spent a year researching a topic.”
By 2014, Quid’s platform was fully baked, joining a cadre of firms that were attracting interest from investors and business leaders who recognized AI’s potential to inform a wide range of executive-level decisions. Among the earliest companies to embraced Quid’s AI-driven insights? The world’s largest retailer, Walmart.
And the excitement hasn’t waned. A recent PwC study found that 72 percent of business executives feel that AI represents “the business advantage of the future.”
What Do Shoppers Really Want?
For Walmart, chief among the insights that AI promises to deliver is a timely snapshot of its customer base’s ever-evolving priorities.
Case in point: In 2017, looking to more deeply understand the 140 million Americans who shop weekly at Walmart, the retail giant tasked Quid with shedding light on how American family life is changing. The resulting report, “The American Family Today,” identified five major shifts that families are facing, all surfaced by Quid’s machine-driven analysis of mountains of content.
It’s easy to see how these trends might be monetized by a company like Walmart. Increased consumer interest in redecorating, for example, might lead Walmart to boost shelf space devoted to home furnishings or hardware. Sure enough, Walmart says it has seen a seven percent increase in sales of tools and other home improvement hardware in the last year, and a four percent increase in sales of home décor items.
Other trends appear more nebulous at first. “Too busy to get sick” reflects consumer sentiment that the cost of taking time off from work due to illness (or to care for a sick child) has become incredibly steep.
Aaron Bernstein, Walmart’s Senior Director of Insights and Advocacy, says, “We saw in our data that there was a huge increase in over-the-counter pharmaceutical sales, sales of vitamins and even sales of garlic and turmeric [which purportedly have healing properties]. People will do anything not to have to go to the doctor, and pharmacists are replacing general practitioners as the first line of defense.”
The Science—and Art—of AI
Bernstein, who spearheads the use of AI tools at Walmart, notes that the report is just the tip of the iceberg.
“Our scale at Walmart is massive,” he says. “Think about how much information exists externally that has relevance to us. We touch on almost every facet of life, so the ability to quickly process information at a high volume is the big value add in how these tools are helping us get to insights and make decisions faster.”
AI is an art form. We're taking and applying science and machine-learning capabilities and marrying it with art.
— Aaron Bernstein, Senior Director of Insights and Advocacy, Walmart
At Walmart, Bernstein adds, “AI is an art form. We’re taking and applying science and machine-learning capabilities and marrying it with art.”
The breadth and depth of Walmart’s current embrace of AI tools is staggering. To date, Walmart has built more than 1,300 networks within Quid, each tasked with investigating a single topic. Those 1,300 networks have analyzed more than 13 million articles of content. Collectively, the data would have taken a human 77 years to read—literally a lifetime of data analyzed in about 3,000 hours of computing time.
One Million New Documents per Day
Dealing with that crush of data is one of Quid’s biggest challenges, says Goodson. “We now have more than one billion documents and are growing at a rate of one million-plus per day, making it one of the largest document storage initiatives anywhere.”
When Quid analyzes those documents, the challenge is exacerbated by the massive storage and data-processing power required. The only viable solution to both of those issues has been the cloud. On-premises systems can’t efficiently scale to Quid’s needs, so today its backend is 100 percent cloud-based.
Despite AI’s arrival in the corner office, Goodson notes that the technology is still a story in progress. “There's just so much to be done,” he says. “For every hard question that a professional has, there will be an AI partner that exists to read everything and support the answer.”