Photo: Getty Images/jetcityimage
Voice assistant takes orders from McDonald’s drive-thru customers
Visitors to the McDonald’s drive-thru might soon be talking to a robot instead of a person when they place their orders.
McDonald’s is piloting a voice-ordering solution at drive-thrus in 10 locations in the Chicago area, according to a CNBC report. The solution is approximately 85 percent accurate, leaving about a fifth of orders to be taken by human staff. While it may take a number of years to implement chain-wide, given the complexities of handling promotions, menu differences and recognition of regional dialects, the technology could represent the next step in fast food automation.
McDonald’s is not the only fast food chain experimenting with automated voice ordering.
White Castle has been piloting a similar solution, according to CNN, but the chain’s voice ordering bot also folds in predictive suggestions. The company plans to enable personalized ordering via license plate recognition, as well as features that take into consideration weather and day-part.
Other QSRs have taken different technological tacks to speed along the drive-thru experience. For instance, mom-and-pop restaurant Fair Oaks in California is using facial recognition to speed payment transactions.
Automation in fast food is a perennially controversial topic. Advocates see it as a way to free up workers to undertake more fulfilling tasks, while opponents see tactics that cut down on labor costs by replacing low-skilled workers with technology.
McDonald’s has already taken many steps to automate portions of its operations previous to this latest pilot. The 2016 launch of its “store of the future” was followed by the chain-wide adoption of in-store touchscreen ordering kiosks. The chain has also piloted other AI-based drive-thru technologies, such as touchscreen ordering that does not require human interaction to place the order.
Voice recognition technology is causing controversy in other areas, as well. A patent by Spotify for voice recognition technology that would allow an AI to gauge a user’s age, dialect, mood/emotions and other characteristics has been decried as potentially dangerous, according to Opus Research. Critics fear such technology could be used to manipulate users.
- McDonald’s is testing automated drive-thru ordering at 10 Chicago restaurants – CNBC
- McDonald’s and other chains are giving their drive-thrus the Jetsons treatment – CNN
- McDonald’s drive-thru AI knows what you want before you order – RetailWire
- Accusations of “Controversial Speech Recognition” Will Chill Introduction of “Voice First” Services – Opus Research
Discussion Questions
DISCUSSION QUESTIONS: Do you see voice-ordering drive-thru technology, once the kinks are out, being widely used in drive-thru operations? Are there any downsides into its implementation at scale? In what creative ways might QSRs consider mobilizing drive-thru staff if the need to take orders manually is no longer present?
Poll
BrainTrust
Christine Russo
Principal, Retail Creative and Consulting Agency
Karen Wong
Co-Founder & CEO, TakuLabs Ltd.
Google Voice, Alexa and Siri have accelerated the consumer adoption of voice assistants and improved the experience. Voice-ordering will become a common way for restaurants to process drive-thru orders as the technology continues to improve. With the current challenges of hiring and retaining employees, voice assistant technology will be considered for any roles that are structured.
Being involved with Theatro – a company that relies on voice activation and voice recognition as the differentiator of the product they sell – I can state unequivocally that this technology has a lot of runway, and QSRs are an ideal place to implement it. The technology will be refined to more than 95 percent accuracy and it all will work just fine. Whether it will be one or four years until full adoption is hard to predict but it will happen in that range.
Hoo boy, here we go again. Are customers going to be happy talking to a machine, rather than an immigrant who really needs that job? I can’t think of anything good to say about this misguided and cynical idea.
AI voice interface is here to stay. Why? It’s cheaper for the seller. Automated retail is not going away. I hate interacting with a fake voice, but I’m probably a dying breed. Younger consumers are much more accepting of Siri, Alexa and the gang.
Voice-operated technology has had a slow (and somewhat bumpy start) start but is improving and will be everywhere. AI voice monitoring is prevalent in CX: detecting mood and “happiness” during the conversation. There are many tech companies providing this AI service that produces “actionable” data. For McDonald’s, this isn’t a conversation of IF voice-ordering should be used extensively; it’s a matter of WHEN.
70 percent of McDonald’s revenue is from drive-thru – so the efficiency and accuracy of their windows are absolutely critical. If fast-food consumers realize value by getting faster throughput, I don’t believe they will care whether it’s a live person or a bot.
I have, as I am sure we all have, dealt with voice-activated responses that were bizarrely frustrating (“I want to speak to a representative!”). But, I recently dealt with one that I didn’t know was a robot. I was astounded and easily half way through the conversation before I realized it wasn’t a person on the other end. A good half of my writing today is done by voice, which gets it better than my typing.
The ability of voice to do the job, whether taking orders for McDonald’s or driving a car, or answer my complex search questions is accelerating at a extraordinary rate. Maybe I shouldn’t say “extraordinary” because it is only following the development path of most digital technology.
Eighty-five percent accuracy will not cut it and I assume that people will be even less likely to cut a machine a break for problems ordering than they would a human on the other side who actually may be offended by unkind words. Can’t you hear it now, a customer becomes upset and swears at the voice-tech that responds that it either doesn’t understand the question or says that it can not answer the expletives directed at it.
This. I have issues with Alexa and Siri with basic requests and I can see this being a problem that would turn me away from going to McDonald’s completely – especially if I have had a bad day and have to repeat my order five times.
Yes, 85 percent won’t cut it. And that other 15 percent has driven me to loud expletives. But 85 percent is not the target of the technology. It is 99.9 percent or better and it will get there.
NLU accuracy from the major providers is in the 95 percent range, which is better than human conversational understanding and more than sufficient. And proper systems are designed with a fallback to humans. Why McDonald’s has such a low understanding threshold is unknown based on details here (typically ambient sound is a big factor), so McDonald’s is no threshold to judge the technology overall.
So right. Too often times I get more frustrated with human responses than AI.
The key phrase is “once the kinks are out.” It’s particularly challenging when you are outside at the drive-thru and road noise creates interference. That said, once the technology is perfected, how would I know that I’m talking to a person or a machine at McDonald’s? It’s not like orders get complicated. And this will be more attractive to the restaurant as labor costs rise.
Yes, I see voice-ordering drive-thru technology being widely used. Why wouldn’t it be? Multi-unit chains (except Starbucks), rarely have associates creating any personal connection with the guest. Starbucks is the only drive-thru where people actually say anything more than “here’s your order.” As labor costs rise and profit pressures on these public companies grow, you’ll see more adoption of technologies like these.
The downside is job erosion but I don’t think that is avoidable in our society.
I’m not sure creative ways to mobilize drive-thru staff make sense, they’d probably cut drive-thru staff hours and use the remaining for QC. This all seems aligned with McDonald’s brand. It’s a very transactional business and I don’t think customers mind as long as they get cheap, tasty food.
This is an exciting time for many businesses! There are kinks to be worked out, but I see this making the operation more efficient. And in addition to the voice recognition mentioned in the article, as the technology advances, facial recognition will help make the experience better. The computer will detect returning customers, make appropriate suggestions based on past orders, and more.
One of the key metrics for QSR drive-thrus is speed of service and another is order accuracy. Unfortunately, we have seen a rise in instances of violence when one or the other does not meet a customer’s expectations. Hopefully McDonald’s will be able to get the order accuracy above 95 percent. To do that will require the system’s ability to handle the complexity of the differences across the items in the network, regional specials, and the ability to understand the difference between a Texas drawl and a Boston accent.
Voice and textual chat interfaces are powered by a form of AI called NLP and NLU. When the set of subject domains is limited to a minimum, it allows the NLU to be more effective, whereby its underlying training and machine learning can be narrow and deep, versus trying to understand many things.
Drive-thrus lend themself to effective NLU and voice interface because the scope is limited. So rather than the variability of an associate that may be distracted, multitasking, poorly trained, or even not customer focused as the point of interaction customers interface with, an AI voice assistant is completely focused, consistent, and improves continually. It’s the one fast-food role where I see no doubt that AI should get the job over humans. I would however suggest having the most cheerful employee passing the food to the customer to end the experience with a happy humanized touch.
I am fascinated by the evolution of how people get their coffee in the morning, and how they get their burgers during the day! I’m sure that consumers of all kinds will be communicating with AI enabled systems and robots on a regular basis at some point in the future. There is so much effort, energy and resource going into creating these kinds of solutions that I have no doubt that it will become mainstream soon. A few years ago there was no “mobile order ahead” from the coffee or donut shop, now it is the only way many people get their morning Jo. I fully expect that McDonald’s will have an app integrated into cars’ infotainment systems where the driver will build the order on the way to the restaurant through interactive voice technology, and will skip the whole drive-thru line. McDonald’s will be serving its “billions” in lines inside the store, through mobile order-ahead (walk up and drive up), through voice enabled drive-thrus, through online meal delivery, and many other to-be-developed methods. The inside operations of the restaurants will need to become hyper efficient as fewer people will be required for customer facing interactions, and higher volumes of orders are coming in.
Voice ordering is simply one of many technology driven changes that QSRs and others are testing to enhance the customer’s experience. While speed of ordering may be an objective, with the right technology improved order accuracy can be achieved. Such technology provides an opportunity for staff to concentrate on the execution and delivery of the order.
“I’m sorry. I didn’t understand. Please listen to your choices.” Just kidding. I think I may hold the record for being the longest-running advocate of interactive voice technology on RetailWire. I became a convert decades ago when I was an early adaptor of Dragon, possibly one of the clunkiest interfaces ever created. So, in a rare nod to consistency, I see voice-ordering and interactive voice technologies being widely used in every aspect of our life — including QSR drive thrus — although, full disclosure, I rarely ever eat any QSR food and never go through drive thrus. As to the downsides, they are obvious. Customer frustration at misinterpreted orders (a problem with person-to-person communications as well) and job loss from technological disintermediation. As to the last question, let’s face it, these workers are not likely to be repurposed.
Having friends with startups in this space and working with QSRs, this type of tech has a lot of potential as long as it’s implemented into a process with choice – e.g. does not exclude those we are not comfortable with voice tech. There’s no turning back time with digitization – yes, high-touch interaction is great but it’s not consistent, especially during peak periods at a QSR which is when drive-thru tech like this really shines.
Even with the kinks, this will be a hit provided there is some video (digital signage) confirming purchases for the drive thru customer. Banks, customer service centers and more have been using voice activated and language processing IVR tech for decades and it’s getting smarter and smoother. The nuances of deploying on scale is one issue, especially with different menu items — but these can also be resolved locally. The need for escalations will still be there, so former QSR staff who asked the questions can service the customer, manage payments more efficiently, and even deliver products to the customer at the window. Depending on system costs, there is the potential labor savings, however many such technologies are still relatively higher cost products so staff replacement will take time as many QSRs will be reluctant to move to the model if the $12-$15 intercom handler is just as easy to find.
Of course this will work. This type of specialized AI voice computing tech is approaching commercial readiness and the cost is likely plunge as it is perfected. Combined with mobile apps already spreading in popularity, this will eliminate one point of friction in the drive-thru experience. Eventually. Just be prepared for the memes to come when amusing and annoying fails happen in the shakedown phase.
In many ways, this just makes sense: a voice-dominated service like drive-thru leveraging voice technology. This seems to check all the right boxes.
However, while we acknowledge that “kinks” need to be worked out, I wonder if it’s those very kinks that will prevent voice from reaching its full potential in QSR drive-thru.
Human beings are fickle. They want to order what they want, they expect you to understand what they want (even if it’s not on the menu) and they are easily frustrated when these wants are not met while sitting in their car speaking into a black box. Sometimes, it’s the human understanding on the other side of that black box that turns these frustrating moments into satisfied customers. It’s hard to replace this connection with technology.
That being said, QSRs should forge ahead with voice technology where it makes sense. This is the next generation of drive-thru ordering. However, the humans making these decisions should remember why consumers visit their restaurants. Even drive-thru has an aspired experience that can be enhanced or diminished with or without human contact. Finding balance will be key.
The question is not when, the question is, how soon and how fast? Once the kinks are worked out, this is a real win for the restaurant industry, where drive-thru and online ordering have become the new norm.
With the pressure on wages and the lack of available workers, this becomes not a nice to have but a must-have if restaurants want to stay competitive.
Just look at some of the benefits:
Voice assistants at drive throughs is a terrific way to extend the consistency of QSRs to the drive-thru. Today, your drive-thru experience depends on who is manning the headset. That introduces a lot of room for variance. Not all of it good. And as noted elsewhere, people have been trained to speak to devices. Take advantage of that training and deliver a high quality experience every time.