By Yixuan Xie
Medill Reports
The next time when you shout into the ordering box at a drive-through, you might not be talking to a human but an automated order taker.
Drive-through windows make up about 70 percent of fast-food chains’ sales, according to a recent study from QSR Magazine. But employees often consider this work as the most taxing job for employees inside a fast-food restaurant.
They are taking orders from multiple lanes. They are processing credit cards, cash or scanning loyalty cards. They are putting food into bags. The next thing you know—the soda cup is overflowing.
If all these factors are not enough to stress out employees, the bevy of new tech tools such as mobile ordering apps and self-service kiosks—complicated and often out-of-order—add more frustration, remember that these are often minimum-wage jobs.
According to the 2018 MIT Technology Review, the work force turnover in the fast-food industry jumped to a record high of 150 percent, which means that a restaurant employing 20 workers would go through 30 in the span of a year.
Since restaurants are either understaffed or full of new hires who are not fully trained, wait times increase. QSR magazine reported that the average drive-through time among major brands in 2018 was 234 seconds, compared with 225 seconds a year earlier.
To solve issues of workforce shortage and slowed down service time, three AI companies are working on creating AI-powered voice assistant for order taking. Let’s take a look.
Here’s why these companies believe that AI voice ordering assistant will be the game-changer of the next generation drive-throughs.
Seconds matter in fast-food drive-throughs. Potential customers are likely to drive away when there is a long line of waiting vehicles.
Valyant AI piloted a voice-based assistant to take breakfast orders at one drive-through restaurant – Good Times Burgers & Frozen Custard in Denver, Colorado – starting in the fall of 2018. The eatery is seeing a 10 to 25 percent reduction in average wait time, with a few orders hitting 50 percent wait time reductions, according to chief executive and founder Rob Carpenter.
“While our AI is carrying on a conversation with the customer, the employees are listening to the exchange and actually preparing the food,” Carpenter said. “So we are seeing a lot of situations that by the time the driver leaves the drive-thru speaker and pulls up to the window, the employee is already walking over with their food ready to hand over.”
Given the fact that customers want to get their food as quickly as they can and get out of the drive-throughs, Carpenter believes that it is a very viable possibility for AI to take over the majority of customer orders within the next 5 years.
“If you’re saying it’s going to take 50 seconds to get your food using the AI and 100 seconds to get your food using an employee, I think you’ll see a lot of people opting for the AI system provided the system is accurate,” he said.
Although QSR Magazine reported that order accuracy rose slightly from 89.4 percent in 2017 to 89.7 percent in 2018, for people who have strong food allergies, any inaccuracy is a life-and-death issue.
That’s the initial reason why Encounter AI came up with the idea of bringing a voice-enabled AI assistant to fast-food drive-through ordering systems. Its technology is currently testing at several drive-through restaurants in the Midwest and on the west coast including a hand full of the top 20 U.S. fast-food brands to handle all three day-parts, said Derrick Johnson, the company’s CEO and former Franchisee and QSR consultant. Due to non-disclosure agreements, he declined to name any of these restaurants specifically.
“I have a friend with a fish allergy. She was like, ‘Hey, don’t cook my chicken sandwich in the same vat of oil that you just cooked the fish.’ The order associate inadvertently didn’t put her order in and it was sending her into very severe allergic reactions,” Johnson said. “So from a voice recognition perspective, we thought that we could precisely understand that order, get that over to the point of sale system and then put an emphasis on the fact that this is an allergy order.”
Johnson declined to provide current statistics on accuracy of its AI voice-ordering platform. But he said accuracy started with at about 85 percent and is continuously improving over time through the process of machine learning.
“Today we handle a normal order at the efficiency of an ordering associate but are monitoring exceptions. So if someone comes in and says ‘I want two bottom buns with my order’ and our system has never heard that before. We want to give it time to learn more of these edge cases,” Johnson said.
Messy language, accents or improper grammar can also result in wrong orders. But it won’t be a problem anymore with the conversational AI technology developed by Clinc, which is currently in late-stage talks with a number of top quick-service restaurant companies in North America to augment the voice control capabilities of drive-thru windows.
Unlike many other voice-powered AI tools, Clinc’s platform is purely data-driven and more flexible, learning from the different ways that people order, according to Dennis Zender, executive vice president of strategy and innovation at Clinc.
“We don’t have decision trees. We don’t have a very scripted approach. We don’t tie into keywords,” Zender said. “We look at sentence structure and have a very advanced setup in order to understand exactly what it is that they are ordering and be able to return a very dynamic response.”
With more orders collected and more conversations with customers, artificial intelligence enhanced by machine learning can remember customers preferences and make the best decision about which food to promote in future visits. Encounter AI is currently using this technology to avoid bombarding customers with items regardless of their preferences.
“What we’re doing is essentially creating a customer profile,” Johnson said. “Last time a customer ordered a chicken sandwich, onion rings, and a shake and this time she didn’t order a shake. That type of personalization and preference is what we can bring to the edge and we believe that’s where the industry is going.”
In addition to previously placed order, Clinic’s AI platform also remembers customers’ favorite orders to optimize the up-sell opportunities.
“Hey, would you like to place your favorite order or would you like to repeat your last one,” Zender said. “We have the ability to bring that forward in the order process as well.”
Besides customized up-sell, AI-powered voice assistant helps drive-throughs become more profitable through labor reduction. Johnson said drive-through traffic has increased approximately 30 percent and, often times, fewer employees are required for drive-throughs that are partnering with Encounter AI.
“We were talking to a franchisee and he is doing $400,000 in sales annually, which is not profitable,” Johnson said. “His choice is either labor reduction or closing its doors. What we’re trying to do is to fix those businesses that are small and give them a fighting chance.”
- Consistent and enhanced customer experience
Having a pleasurable service experience at drive-throughs often amounts to how nice the employee is who takes the order. AI voice-ordering assistants will never sound annoyed, unhappy or tired, contributing to a more consistent customer experience.
And AI companies are making other efforts to help customer experience become more delightful such as Encounter AI’s emotional recognition from voice.
“Is this person happy? Is this person sad? Just first principle customer service nuances like that,”Johnson said. “If this person is speaking in a dull voice, we know that we probably want to come back with more enthusiasm.”
In addition to that, the voice assistant can also adjust its response to customers based on the expressiveness in a customers’ voice. It can sound excited, calm or inquisitive to best match customers’ emotions, creating a positive experience for them.
Despite many concerns that AI will harm the relationship between customers and restaurants due to its lack of human involvement, Valyant AI has seen positive customer adoption with moments of customer delight. It’s been a welcome surprise to see customers saying “thank you” even though they know they are talking to AI. In those situations, Valyant’s system will quickly come back with “you are welcome.”
“Those are so easy to overlook but are really critical to the overall experience of making people comfortable with AI,” Carpenter said. “So the next time those customers come back and they remember the last time AI said, ‘You’re welcome,’ they just tend to humanize it a little bit more, be a little more friendly, easygoing, and patient when interacting with the system.”
Despite all the potential benefits advertised by these AI companies, experts have their doubts about the trend of replacing human being with virtual voice assistants. Ken Forbus, an AI professor at Northwestern University thought it is a little early.
“For order taking, language technology could be pretty good for that, except that it’s not clear that it is better than humans,” Forbus said.
Forbus pointed out two issues. One is that, when people order food, they might have questions. Anticipating all the kinds of questions customers might ask could be very hard and that’s when natural language is needed. The other thing is that today’s robots haven’t been smart enough and their sensors are not very good so that they can’t really handle things the way human do.
“The question for my mind is that: is automating the ordering process the right thing or is training your employees” a better way to go? Forbus said.
Photo at top: Valyant AI CEO Rob Carpenter demonstrates how the company’s conversational artificial intelligence platform works at Good Times Burger & Frozen Custard restaurant drive-thru in Denver. (Courtesy of Valyant AI)
