It’s not quite the robot overlord takeover as seen on the sci-fi big screen, but the limited-service segment is experiencing an influx of artificial intelligence (AI) as more chains invest in automation technology.
One of the buzzier realms attracting investment from big names like Taco Bell, Wingstop, Pizza Hut, and Burger King is automated chat ordering, which leverages already-popular messaging platforms and voice applications like Slack, Facebook Messenger, Twitter, SMS, Kik, and Amazon Echo.
Chatbots are computer programs that use AI to mimic human conversations. Bot interactions can be simple, like ordering a pizza, or more complex, like troubleshooting a computer problem. The big draw is their ability to condense multiple services into one chat contact—which, in effect, means they can replace individual apps. Rather than closing Facebook Messenger and opening Uber to request a ride, for example, a consumer would need only message Uber.
Messaging apps have become indispensable for today’s smartphone-wielding consumer. As of July, Facebook Messenger had 900 million monthly active users worldwide, while Twitter logged 313 million monthly active users as of Q2. And office productivity app Slack boasted 2.7 million daily active users as of April.
For chains like Dallas-based Wingstop, which recently went live nationwide with a chatbot on Facebook Messenger and Twitter, using platforms that already have high levels of engagement seemed like a smarter bet than banking on every customer to download the Wingstop app on his or her already-crowded phones.
“Messaging platform apps are where the growth is in terms of downloads; single-use platforms are not,” says Stacy Peterson, chief information officer at Wingstop. “Restaurant apps are important, and your most loyal customers will download them, but not everyone is going to, so being able to allow someone to order without committing to downloading your app and creating profile frees up space on their phone. That’s a real value proposition.”
Expanding digital ordering through social media is also a natural fit for the brand, which already engages customers through other social channels.
Wingstop enlisted third-party enterprise conversation platform Conversable to integrate the chatbot with its backend e-commerce system. The AI computer program adapts the menu and pricing into conversational language for Facebook Messenger and Twitter, controls the conversation, and flows final orders into the restaurant’s POS system.
“A restaurant doesn’t have to be a digital company [to deploy this technology]—just have the ability to integrate with legacy enterprise systems such as PCI-compliant POS systems [customer relationship management systems] like Salesforce or a loyalty provider,” says Ben Lamm, CEO and cofounder of Conversable, which partnered with technology services company Mindtree in September to help standardize backend integration of its platform.
Lamm points out that chatbots not only cost chains far less than phone calls or call centers, but their conversational engagement also lends itself to a simplified path to ordering. While online and mobile app ordering is fairly consistent, it doesn’t allow customers to skip steps in the process or order as organically as they would with a server, he adds.
Taco Bell teamed up with office productivity app Slack to develop its own Siri-like chatbot this spring. With the help of its digital agency Deutsch, the taco chain leveraged Slack’s open-architecture platform to put its own, irreverent spin on TacoBot, which is still in beta.
“Early on it was pretty mechanical,” says Andy McCraw, Taco Bell’s manager of e-commerce innovation. “You had to know the special set of forward-slash commands common to Slack to interact, so it was targeted toward more of a techy user. Once we determined we could start to leverage the AI to build out a natural language ordering interface, then all of a sudden it got interesting.”
TacoBot not only engages users conversationally—“sounds good; do you want to keep adding stuff to your order?”—but it also cracks jokes, uses emojis, answers trivia questions, and even adds a cup of water to an order if the customer mentions being hungover.
The biggest technological challenge for both Taco Bell and Wingstop thus far has been working within the functional constraints of each app.
Facebook has a much richer interface with pictures, while Twitter is fairly limited to A, B, and C-type options via text.
Taco Bell had to adjust to similar guardrails with Slack’s text-heavy format, but the chain is building TacoBot to be “platform-agnostic,” so it can connect to the different channels its customers are using, including its own mobile app and e-commerce site.
“We don’t want to rebuild from scratch and retrain the AI in every new channel,” McCraw says. “We also don’t want to dictate for users where they have to be to interact with Taco Bell. If TacoBot is only on Facebook Messenger, for example, we’ve potentially excluded a huge percentage of our customers. The more relevant and accessible you are, the more your value spreads across customers.”
The other main hurdle at this early stage is spreading awareness about this fun style of ordering.
At Wingstop, digital ordering as a whole is on the rise, accounting for 16.9 percent of total sales in the second quarter (up from 15.8 percent in the previous quarter) and averaging $4 higher per ticket than in-store orders.
“We’ve done that organically—as in, no big events where we gave away product to get people to download it,” Peterson says.
But in the hopes of keeping it more top of mind, she adds, the next step is to have the bot directly connect with those highly engaged customers who are already talking about craving Wingstop on social channels. Looking ahead, the chain is also exploring SMS and voice channels for car ordering as those technologies gain traction.
Because chatbots are fairly limited in the type of impact they can have, Taco Bell’s eye is also on the future—namely how AI can help extend the brand’s personality and helpfulness beyond the droll chatbot into resolving consumer pain points that digital interactions have fallen short on in the past.
“I think if all you’re going after is a chatbot, then you’re missing the really big opportunity,” McCraw says, adding that AI advancements should strive to improve the overall customer experience. “That is a unique type of interaction that’s really hard to do in other traditional channels.”