Siri and Alexa have certainly gotten people accustomed to using artificial intelligence on a daily basis, and according to Gartner research, AI will indeed amount for 85% of customer relationships by 2020. Retailers and brands are increasingly using AI in the form of chatbots to interact directly with consumers, and while it’s still early in the game, what’s happening in this dynamic space is worth a closer look.
What is a message chatbot exactly? Simplified, it’s the AI version of live chat on a message app, where a smart “bot” (rather than a human) answers your questions in real time. Popular platforms for chatbots are Facebook Messenger, Kik and WhatsApp, meaning people can interact with the bots without having to download a separate retailer app.
This taps into a shifting tide with lots of e-commerce potential. In 2016, 1.6 billion people used mobile messaging apps, according to Forbes, and that number is expected to reach 2 billion people, or 80% of all smartphone users, in 2018.
At the simplest level, chatbots can help a user better find what she’s searching for. One level up, “fashion savvy” chatbots can provide suggestions, add-ons and one-on-one styling help that previously only existed in a boutique or retail store. Chatbots also listen and learn, much like an actual salesperson, and grow “smarter” with each interaction.
Headliner Labs is a female-owned development company that has created chatbots for fashion retailers like jewelry company Jemma Wynne and scarf company Donni Charms. The team of sisters now apply their respective experience from Fab.com and government cybersecurity to artificial intelligence in the retail, media and hospitality industries, all set to improve the user experience.
Headliner Labs calls its chatbots “a one-on-one conversational interface between customers and brands powered by AI and learned responses.”
Interfacing via messenger apps keeps things in one spot and means less “friction” at the retail experience. “The idea is that customers are already on Facebook and Messenger: 71% of the time spent online is on mobile, and when you’re on mobile, you’re usually messenging. This way they can interact with the retailer without downloading a new app,” says Caroline Stern Klatt. “People are on Facebook all day long, but even a retailer’s ad on Facebook asks consumers to click off the page. Message bots let you message that retailer right from there.”
How Humanlike Should They Be?
While a chatty manner makes chatbot exchanges feel more natural and organic, should a chatbot try to fool consumers into thinking they’re interacting with a real human? There is debate on this subject, and indeed many computers have been known to pass Alan Turing’s notorious Turing test (the upcoming remake of Blade Runner also brings up the “replicant” conundrum). But when it comes to customer service, consensus is that transparency is best. According to a CEB study of thousands of people, transparency ranked highest priority for consumers with the largest impact on loyalty, according to Astute Solutions.
That said, some chatbots will try to seem humanlike, right down to colloquial chattiness or by showing the well-known “…” ellipses when they are ostensibly “thinking and typing.” This is a quaint touch, as the bot surely knows the answer instantaneously. Other “human” touches include emojis like thumbs-up symbols and smiley faces.
One of the concepts of chatbots is that instead of searching through a site’s numerous offerings or filtering by keywords, a user can actually tell a chatbot what she’s looking for or ask for help and suggestions. In essence, the way one would interact with a salesperson/stylist in a boutique.
“The magic of in-store retail is lost online and we’re trying to bring that back by helping identify what’s best for the customer,” says Headliner Labs’ Dana Stern Gibber. “We start with some multiple choice questions to get a sense of the customer’s style, and have them type in some open-ended answers too.”
Once consumers opt in, the bots can push out add-ons that can accessorize an earlier sale, or just things that go with that style. And while retailers can do this via email, there is usually a donotreply at the other end. Chatbots will answer you instantly.
Jewelry company Jemma Wynne uses Headliner Labs’ chatbots to help consumers create signature stacks of jewelry, showing them to the “bot” along the way for assistance.
Scarf company Donni Charm recently added chatbots to much success. “I wanted Donni Charm to have a chatbot because I want my brand to be interacting with customers where they spend phone time—in their messaging apps,” says Founder Alyssa Wasko. “Retail companies like ours must innovate while staying true to our brand voice and aesthetic, and because of their background in fashion and commerce, the Headliner Labs team totally got that.”
While consumers can ask a bot questions and initiate a search, it can also go the other way. “Donni Charm sometimes initiates messages when the conversations originate from our website to a chat that happens in Messenger,” says Wasko. Consumers can search the message app the same way they’d search/select a friend to chat with, but Messenger is starting to feature bots and serve them up in search. “This is a great way for new people to find us.”
Again, much like a live salesperson in a store, chatbots can push a customer into larger orders and add-ons. “Our goal was to increase the overall direct-to-consumer sales and get new customers, and we’ve seen major success with those objectives,” says Wasko.
STILL IN THE LEARNING CURVE
But as chatbots rise up (in popularity, not in the take over the world sort of way), they are hardly perfect. Bots have a learning curve just like real customer service agents, and not all can get there fast enough for companies or customers.
Facebook noted a few months ago that it was “refocusing” its use of AI, according to Digiday, after its bots “hit a failure rate of 70%.” To put this in perspective, a 70% failure rate means bots could only manage customer requests 30% of the time without resorting to human intervention.
E-tailer Everlane tried out chatbots as part of Facebook’s Messenger program but confirmed that it decided to revert back to email. Another customer, Spring, also noted that the Facebook Messenger bot “didn’t have the level of personalization people expect.”
Snags can result from the bot not understanding voices, colloquialisms or more. It all comes down to cost analysis. While replacing human chat with a bot chat might save on payroll, what does it cost in frustrated or lost sales?
The answer isn’t known yet, but as more companies try out bots, it will be interesting to see.