The next time you're in a Lowe's hardware store don't be surprised if a robot zooms up to ask if anyone needs help. And when you order your morning joe, a programmed bot may be whipping up that venti skim macchiato at Starbucks.
Watch out, folks, the age of artificial-intelligence-powered retail is upon us, promising an unprecedented amount of data and information that can help merchants grow their businesses. Once the stuff of sci-fi novels, AI is poised to juice up everything from customer service to merchandising—except jobs for humans.
"AI is going to be like electricity or the internet—it's going to be foundational technology [for] which most things are built," says Kyle Nel, executive director of Lowe's Innovation Labs. "It becomes so interwoven into robotics and everything. It takes CRM, it takes all of this other stuff that we've been doing for so long and it makes it better than the sum of the parts."
Such data-heavy tactics are particularly intriguing for retailers, who are already using AI in crazy-cool ways that leave other marketers envious. "These are industries that sit on tremendous pools of data about their customers, but [ecommerce] conversion rates sit around 2 to 3 percent—in the physical retail world that would be considered abject failure," explains Jonathan Epstein, CMO of 9-year-old AI player Sentient.
According to market firm Tractica, global revenue from AI will skyrocket from $643.7 million in 2016 to $36.8 billion by 2025. Also by 2025, Forrester expects that cognitive technology—including AI and automation—will replace 7 percent of U.S. jobs.
It's hard to say at this early stage whether Main Street is ready. Adweek commissioned Survata to ask 1,000 consumers how they feel about artificial intelligence. While 74 percent said that they had positive experiences with AI in retail stores, in another question 65 percent stated that they are not cool with a robot replacing a human worker in stores. At the same time, 57 percent said that they would happily exchange messages with an ecommerce merchant's chatbot online.
Regardless, it's no wonder that every technology and software company claims to have AI expertise these days—although definitions vary wildly and it can be hard to pinpoint what exactly AI is. Expect those lines to blur even more in 2017.
"A lot of companies are going to start using the term 'AI' to position themselves differently in the marketplace," says Jeff Malmad, managing director, head of mobile and Life+ at Mindshare North America. "It becomes an opportunity for them to [sound like they] have an advantage over their competition."
Yet legitimate examples do exist—from store-mapping robots to tech that predicts if a dress will be an instant best-seller before it's even designed. Here are a handful of examples of retailers using AI in revolutionary ways.
The Big Brain: IBM Watson
IBM famously brought AI into the IRL mainstream in 2011 when its supercomputer Watson won $1 million on Jeopardy. Since then, marketers are just starting to get their hands on the souped-up software and its dizzying number of data implications. Watson can process "hundreds of millions of pages" worth of data each second, says Stephen Gold, vp of Watson. Watson turns 8 years old this year, and it is now smart enough to spit out just as much data as it takes in.
"2016 was kind of that pivot point where we saw the output start to equalize with the input," Gold explains. "We can generate a psycholinguistic profile of an individual in literally milliseconds." Such applications can factor in emotion, tone, language, sentiment, purchase history and social media stats.
That's music to data-intensive retailers' ears, especially those who scrutinize such insights before pumping money into faster and better ecommerce sites. Womenswear designer Marchesa, as well as retailers Macy's and The North Face, are tapping into Watson in fascinating ways.
In May, Marchesa designers Georgina Chapman and Keren Craig enlisted Watson to design a dress for the annual Met Gala, crunching hundreds of images to find the right fabric and color. The gown was also embedded with tiny LEDs that changed colors based on real-time Twitter chatter about the event. Based on its success, Gold claims that Watson can now correctly predict success before a stitch is sewn. "As crazy as that sounds, it's feasible," he says. "Can you really anticipate a best-selling dress before you ever design it? The answer is yes."
Early evidence also suggests that AI can boost ecommerce sales. Last December, The North Face started piloting a website that greets coat-seeking shoppers with a series of questions—like where and when they plan to use the coat—to create contextual product recommendations. Shoppers who use the AI tool convert more often than those who do not. And the average order value is "a mid double-digit percentage" better for shoppers who use AI compared to average shoppers, says Cal Bouchard, The North Face's vp of digital commerce and experience.
What's more, The North Face's shoppers are queried after they use the AI tool to learn if they would use it again. Three out of four customers say yes, they would. After a year's worth of testing, Bouchard says she's now looking for ways to build content—like say, a travel guide about the location where someone is going—into the site.
"I think there's a lot more that AI can do than just recommend the next purchase," she says. "I see the future of AI adding content that would make your adventure more interesting. What about product education, content about where you might be going or what activity you might be doing?"
The Trendsetter: Starbucks
The next time you want a triple venti soy no-foam latte, don't be surprised if you place your order by talking to your smartphone. Indeed, typing is so 2014 for digital-savvy Starbucks, and AI-powered, voice-activated apps are the new red-hot piece of technology.
The Seattle-based roaster pioneered mobile payments six years ago with its popular app that rewarded users with club points for buying their daily coffee. Later this year, Starbucks will up the ante on ordering by building a Siri-like virtual assistant into its app. The feature—dubbed My Starbucks Barista—allows users to place their order by tapping a button and talking to the virtual barista. The bot then pings the order to a nearby store where an employee makes the drink. Users can then pick it up within minutes, bypassing stores' long lines to the cash register.
Mobile payments are already a huge deal for Starbucks—25 percent of the chain's U.S. transactions come from a smartphone. Put another way, the stat is mind-boggling: Eight million—or about two-thirds of the chain's loyalty-program members—utilize its app to pay for coffee at least once per month. One-third of those customers also choose to order beforehand and skip the line.
So it's no surprise then that Starbucks' bet on AI is chock-full of data. The app's algorithm tracks what people buy to recommend similar products and offers in the future. Someone who prefers iced to hot coffee could get an coupon for free food (the algorithm detects a 3 p.m. snack), for example, while someone who buys a cup of coffee every morning may be able to play a game within the app (the algorithm detects a long morning commute). All told, AI in theory will allow Starbucks to create a limitless number of offerings.
"It is not a single algorithm that's used across the entire population," Gerri Martin-Flickinger, evp and chief technology officer at Starbucks, told the audience during Starbucks' Investor Day last month. "It's actually a data-driven AI algorithm based on your own preferences, your own behavior as well as behaviors that we're trying to drive."
The Sci-Fi Geek: Lowe’s
Roaming robots will be regular sight in the aisles of home improvement retailer Lowe's. After testing store-mapping robots in a couple of locations for the past two years, the home improvement retailer will unleash its own branded machines called LoweBots in 11 Bay Area stores this year.
Each store is equipped with a couple of LoweBots that use natural-language processing—a core component of AI—and respond to customers' questions like, "Where are the hammers?" The machines also have sensors built into them so that they can detect that a person idly standing in front of the lightbulb section, for instance, may need help. "It can tell the difference between a person and an object, so it will engage a person and talk to them," says Lowe's Innovation Labs' Kyle Nel. Meanwhile, mapping technology runs constantly in the background to keep tabs on inventory, which turns into stats that determine sales patterns.
"It's learning things that we never knew before," Nel adds. In one case, he says, store-level data show "what is happening at 3 o'clock on a Tuesday" in a specific store.
Lowe's is also working with Microsoft to launch AI-powered virtual reality in five stores during the first quarter that allow shoppers to put on a HoloLens headset and digitally design a kitchen. Using Microsoft's AI platform Cortana, the experience will scrape data from a user's Pinterest account to serve up recommended products.
"Instead of showing you 1,000 different types of curtains, we can show you a bunch that we think you'll like based on your style profile that you've already created on Pinterest," Nel explains. In other words, deep algorithms are about to become the new interior designer. "Over time, it just gets smarter to the point where it feels like, 'Oh my goodness is this my style curator or an interior designer?'"
The Popular Kid: Amazon Echo
Perhaps no other company on Earth has the kind of deep ecommerce algorithms and pipes that Amazon has, so it's little wonder the firm is ahead of competitors Google and Apple in making the AI-powered smart home popular. "People very quickly understand the power of AI in retail because of Amazon," asserts Andy Hood, AKQA's head of emerging technologies.
Amazon Echo is the tech giant's latest brainchild and runs on an AI platform called Alexa. According to investment firm Mizuho, Amazon will sell 41.3 million of the cylinder-shaped units by 2020, and retailers including General Electric, 1-800-Flowers and P&G's Tide are already lining up to create branded skills that prompt consumers to buy their products from the comfort of their couch.
GE is even working to embed Amazon's AI into its own products. The industrial behemoth will start selling a branded LED lamp with Alexa built in this year that allows consumers who don't own an Echo to ask questions and access information. The brand is also working with Amazon to sell a line of science kits called Labracadabra that are used to create simple experiments—think miniature volcanoes made with lemons and baking soda. After buying a kit of equipment online, a branded Echo "skill"—which is a voice-recognition-enabled feature in Amazon vernacular—can read the instructions out loud and walk folks through the steps of the experiment. "As smart-home technology and AI-boosted voice assistants continue to make their way into homes across the country, the Labracadabra skill is another touch point that will make these science experiments fun, easy and educational," explains Sydney Williams, global digital and social media marketing manager at GE. "The skill will learn and adapt based on how people are responding and interacting to the experiments."
Echo is just the tip of the iceberg when it comes to Amazon's futuristic retail endeavors. Last month, the ecommerce giant opened an 1,800-square-foot grocery store called Amazon Go in Seattle that doesn't have checkouts. Instead, people download an AI-bolstered mobile app that uses locational sensors to track which items someone picks up and leaves the store with. Shoppers simply walk out of the store and the app charges the amount of the groceries to their Amazon account. While only Amazon employees can currently use the store, the company plans to open it to the public in 2017.
"The data is having this profound effect on the amount of things that you buy and the things that you're exposed to that you might buy," AKQA's Hood says. "Taking that off a website and being able to do that in a physical store is a very powerful idea."
The Data Nerd: Cosabella
In an age of mobile and social shopping dominated by the likes of Instagram and Pinterest, it may seem illogical for retailers to nitpick over the small details of their own ecommerce sites. But for Italian lingerie brand Cosabella, digital is an irreplaceable avenue to sell directly to consumers. And the Miami-based team is obsessed over the nitty-gritty details of its website, which requires constant A/B testing and attention.
That need led to Cosabella working with AI player Sentient to take some of the heavy lifting off its plate this fall, employing technology that matches shoppers in real time with website designs they're most likely to engage with and, ultimately, respond to.
"This allowed us to condense six months to a year's worth of testing into 30 days," notes Courtney Connell, Cosabella's marketing director. "It's almost like you have this digital expert there who does that work for you and reduces the hours I personally would have had to put in."
Here's how it works: AI technology digs into small details of the brand's website such as the color of buttons, image size and copy to determine which areas are most effective, then changes them on the fly. For example, switching the color of a button on a checkout page to pink increased conversions for a product by 34.9 percent. All told, AI has boosted Cosabella's sales by 35.6 percent.
The retailer is currently testing the email sign-up box on its websites as either a pop-up or push-down message, and plans to use AI to play with the design of its category and product pages. These are old internet retail tricks that have always depended on a certain level of gut instincts to start out a test—AI simply removes the human guesswork.
Connell says that AI actually frees up her team to focus on bigger projects. Graphic designers, for example, can test thousands of versions of creative instead of creating a handful of ideas.
"It's changing my role to be able to have more strategic, high-level input and not have to be so caught up in the day-to-day amount of time," she adds.
This story first appeared in the January 2, 2017 issue of Adweek magazine.
Click here to subscribe.