The battle for your online shopping dollars has largely been waged on websites and, more recently, smartphone apps. Now, retailers are looking to another digital tool to win your money and your loyalty: An army of chatbots.
Chatbots — the name for robots that simulate human conversation — have been thrust into the spotlight in recent weeks amid a flurry of new experiments in how they might be used to shape the future of shopping. Retail heavyweights Sephora and H&M recently launched bots on messaging app Kik that help shoppers browse and buy their products. Taco Bell showed off its TacoBot, a way to use the messaging app Slack to place a meal order. And on Tuesday, Facebook announced it has created a platform that allows companies to develop bots that run within its Messenger app, which has some 900 million users worldwide.
For many of us who’ve used bot-powered digital assistants such as Apple’s Siri or Microsoft’s Cortana, bots may seem like a novelty. You may find yourself laughing at their misunderstandings more often than you’ve found yourself appreciating their utility.
But evangelists of the technology say that bots are poised to be at the center of a crucial paradigm shift in how we think about using the Internet. While a Web browser might once have been our front door to the Internet and apps often play that role today, experts say that bots could soon become our primary digital gateway. At a conference last month, Microsoft chief executive Satya Nadella said, “Bots are the new apps.”
The case for a bot-centric future goes like this: Smartphone users have proved they are only willing to download and spend time in a limited number of apps. So companies might be better off trying to connect with consumers in the apps where they are already spending plenty of time. And proponents say that a bot can potentially provide greater convenience than apps and Web searches because it can understand natural speech patterns.
Because bots are designed for one-to-one conversation, they may ultimately find their most logical home in messaging apps, which are seeing explosive growth in users and are the digital-communication channel of choice for Generation Z. There were 76 million people in the United States using messaging apps such as Kik, WhatsApp, Facebook Messenger in 2014, according to research firm eMarketer. That figure shot up to 113 million last year, and is it expected to surge to 177 million by 2019.
It is against that backdrop that big retailers and Silicon Valley are racing to develop ways to use bots within messaging apps to deliver customer service or to enable browsing and buying.
When you start a conversation with H&M’s bot on Kik, for example, it kicks things off by trying to get a sense of your personal style. It shows you three pairs of outfits, asking you to decide in each set which look you like better. From there, users can go back and forth with the bot asking it to show different outfits centered around a certain type of item: If you input “lace dress,” the bot might show a complete look featuring such a dress along with a jean jacket, silver bracelet, and ballet flats. Customers can tap the image of each item to purchase it. The bot’s chatter is peppered with emoji and slang like “outfit inspo” and “Perf!,” an effort to make it feel like an exchange you’d have with a friend.
In retail industry jargon, this is coming to be known as “conversational commerce,” and brands are betting on it because of some distinct advantages it could provide in connecting with shoppers. As Kik spokesman Rod McLeod points out, when shoppers engage with a bot, they are almost by definition in a different frame of mind than someone who simply sees a retailer’s display ad on a website.
“The user has to opt into the conversation. So that’s kind of an interest indicator from the get-go,” McLeod said.
And if companies are able to nail the bot’s language-processing capabilities, it could potentially create new convenience for the shopper.
“The web or the app world, they forced humans to behave like computers,” said Beerud Sheth, chief executive officer at Gupshup, a company that offers a bot-building platform. “In messaging, computers have to behave like humans.”
Chatbots, to be sure, are not new to the technology landscape. Apple’s Siri and Microsoft’s Cortana are powered by bots, and bots have long been in use for things like trying booking travel arrangements or getting information from your health insurance provider. But we appear to be on the cusp of a big burst of bot innovation, with experiments underway for using them not only for shopping, but for news, banking and hailing rides. Microsoft is so bullish about the future of bots that it recently debuted a new suite of tools to help developers build them. Facebook, too, is betting that bots will become a crucial way for companies to interact with their customers in the fast-growing Messenger app.
In part, this might be because companies believe that products like Siri and Amazon’s Echo have groomed customers for it.
“I think people are getting used to using this type of conversational interaction to request things,” said Yory Wurmser, retail analyst at eMarketer.
But it may also reflect the fact that recent improvements in natural-language processing and artificial intelligence capabilities are making the possibilities of chatbots more intriguing.
At Taco Bell, for example, such innovations were crucial to development of what would become the TacoBot. Andy McCraw, the chain’s manager of digital innovation, said of an early test iteration, “To be honest, it was a little dry. So the challenge was, how do we inject some personality in this? Natural-language ordering all of a sudden made this way more interesting and a lot more fun.”
In its current form, the TacoBot allows users to place a carryout order for a Taco Bell meal from within Slack, a popular messaging app that is used in many workplaces. It is currently only available to beta testers, but they plan to roll it out for wider use soon.
And yet chatbots are still in their infancy, and that means using them comes with potential stumbling blocks.
Sephora’s Kik app, for example, was able to quickly help me find a Clinique eye shadow when I asked for one. But attempts to be truly conversational with the bot seemed to trip it up. When I asked it to find a nude-colored lipstick and it showed me one I didn’t like, I wrote, “That’s not what I had in mind.” The chatbot then sent me an image and reviews for Bumble and Bumble Color-Minded Shampoo, suggesting it didn’t quite understand me.
And Microsoft recently had a PR disaster with Tay, a bot powered by artificial intelligence that was designed to learn speech patterns from its interactions with users. When Tay began offering up racist statements based on those learnings, Microsoft was forced to temporarily pull Tay from the Web.
“A.I. is all about learning, so it gets better as you do it,” Wurmser said. “But the danger is you’ll drive people off before it’s refined enough.”
Retailers could avoid a Tay-like snafu by configuring bots to learn from only a narrow array of sources, rather than from all the online chatter that comes their way. (And some chatbots don’t rely on artificial intelligence.)
“In terms of Tay, that seemed to be far more strongly linked to the fact that it was exposed and linked to so many parties,” said Ross Rubin, senior director of industry analysis at app analytics firm App Annie.
But the incident still reflects the challenges of taking a chance on emerging technology at such an early stage.
Sucharita Mulpuru, an e-commerce analyst with Forrester Research, said the success of bots might turn out to vary widely across companies and sectors. For example, she sees value in ride-sharing company Uber’s bot to order a car within Facebook.
But, “Does makeup or men’s clothing need to be purchased in this way? I’m less certain it has that much utility,” Mulpuru wrote in an email.