Craig Brabec, the fast-food giant’s chief data analytics officer, discusses how the organization is using AI to boost customer experience and optimize internal operations, among other successes. As one of the most recognizable brands in the world, McDonald’s is always seeking ways to provide faster, better service to its loyal customers—and data and AI may offer an opportunity like none other.
Source: An Appetite for AI at McDonald’s
“The scale of opportunity is tremendous because we serve approximately 65 million customers every day during every part of the day,” says McDonald’s Vice President and Chief Data Analytics Officer Craig Brabec, who has more than 25 years of experience in corporate strategy and data analytics and was previously director of global data insights and analytics at Ford Motor Company. “For instance, we’ve found that customer demand patterns since the COVID-19 pandemic have shifted considerably, including changes in McDelivery and Drive Thru, so we can quickly recognize and make the most of that opportunity through data analytics and AI technologies,” says Brabec, who in his role helps define and infuse data across the global enterprise and establish best-in-class strategy and governance.
In this “AI From the Front Lines” interview, Brabec recently spoke with Beena Ammanath, executive director of the Deloitte AI Institute, about the iconic fast-food brand’s data-driven AI journey and the challenges and myths organizations must overcome on their path to AI success.
Ammanath: Craig, we both use AI with our families in our personal lives. My younger one has entire conversations with voice-activated devices and he’s so polite! But as a business executive, you’ve lived and breathed the AI journey for nearly two decades of your professional life. These days, nearly all organizations use data-driven AI technologies to improve efficiency, while mature adopters are also harnessing it toward boosting differentiation, according to Deloitte’s most recent State of AI in the Enterprise report. What do you see as the biggest business opportunities for AI and data analytics today?
Brabec: I’ll start by agreeing that AI has been transformational in my family, especially for my daughter with special needs. AI has allowed her to access her music independently in a completely frictionless way.
In addition to AI’s personal effects on my family, data-driven AI has the potential to allow McDonald’s to offer a faster and more personalized customer experience. To be honest, speed and convenience have always been in our McDonald’s DNA. We’ve put a lot of work into our delivery and Drive Thru areas of the business and saw some of that pay off during the COVID-19 pandemic.
But the business opportunities for AI and data analytics go beyond customer experience at McDonald’s. We also take a full enterprise view of internal AI opportunities related to supply streamlining, restaurant operations, inventory management, demand planning, and equipment repair before failure, among many others.
Now, as AI use expands, even skilled adopters are recognizing the existence of a “preparedness gap” that can span strategic, operational, and ethical challenges, per the State of AI report. With this in mind, what AI challenges should business leaders consider?
A common challenge is recognizing that we are still in the early stages of AI. It’s not a silver bullet: It requires a high volume of quality data, solid analytics capabilities, and strong business applications. When you adopt AI, you often have to create the data, train models, and do labeling. Sometimes the solutions are still fairly linear and can be brittle and difficult to flex when the models evolve.
It’s also important to note that narrow AI still dominates today’s solutions, which for the most part address very specific problems under a defined set of operating conditions. With new technologies and developments the data science community will be able to go after wider opportunities, but for now, it’s still quite narrowly focused.
At McDonald’s, we’re addressing this preparedness gap with our talent—from how we hire to how we ensure our employees have the tools they need to build up AI acumen within their functions or markets. We are not naïve about the limitations of AI, so it’s important that we think beyond current constraints and continue to innovate to meet our customers’ needs.
That’s very true. According to Deloitte’s 2021 Tech Trends report, many organizations experimenting with AI report feeling hamstrung by clunky, brittle development and deployment processes that can stifle experimentation and hinder collaboration between product teams, operational staff, and data scientists.
Another challenge I see all the time is establishing trust in AI. The increased use of deep learning techniques has caused people to demand explanations of AI models and focus on ethics. People don’t want AI to be a black box. Instead, they really want explainability—to understand what is driving recommendations. That’s a good problem to have, though, because it shows that people see the value in making business decisions driven by data analytics. So the challenge then becomes helping them to understand what is driving those recommendations.
But do you think there can be a playbook when it comes to ethics in AI? After all, explainability may mean different things for different people and different industries. Deloitte’s Trustworthy AI framework, for example, includes six different dimensions to build trust in AI, including assessing whether AI systems are fair and impartial and putting policies in place that clarify who is responsible for the output of AI systems’ decisions.
I think the demand for explainability is a good thing because it ensures we are having a discussion about AI and using it appropriately. But in many ways, trust comes from demystifying things: We’re continuing to educate our professionals and business leaders about what the models are actually doing and how they work. Our team at McDonald’s operates in a hub-and-spoke model; the spokes include our global markets, our various global and market functional departments, and initiatives such as digital, Drive Thru, and delivery.
I’ve seen companies organize their AI teams very differently, depending on their maturity in this area. I wanted to highlight how different companies are succeeding by organizing in a variety of ways, especially if they are just getting started. How are your teams set up at McDonald’s?
I absolutely agree that one size doesn’t fit all. On the one hand, I think understanding how to get AI capabilities to the front lines is essential. On the other hand, there are areas that you want to be strategically aligned and you need a central core. At McDonald’s, our operations are fairly decentralized to help support our franchise model, so we need to be thinking about AI in that way. At the same time, we are starting to expose the broader McDonald’s organization to AI and are working to share this thinking throughout the organization.
Of course, AI challenges can also stem from misaligned expectations, particularly around myths that have been perpetuated about what these technologies can do. What is the biggest myth you have heard about AI?
I think the biggest myth is the idea that AI will follow the traditional hypercycle of technologies and that its development will eventually slow down. Those who might say the next “AI winter” is coming are not typically on the front lines and understanding the transformations taking place.
I see AI gaining traction in every industry, creating value for individuals and society as a whole. There are huge volumes of quality data available to enable AI models, and we at McDonald’s are increasingly smarter about improving business acumen across the organization.
Right. I hear a lot about the “AI winter” in Silicon Valley, in the context of ethics and explainability and transparency—that is, should they even be doing this without having all the guardrails in place? I wonder how it will all shake out. That said, I completely agree that, in the non-Big Tech world, there is tremendous opportunity and transformation. In that regard, what do you wish you had known before you started your AI journey?
I have recognized repeatedly the importance of starting with a clear definition of the central question and the business opportunity. The insights and learnings on the AI journey may cause directional adjustments to that central question as you uncover the truth from the data, but keeping it as your beacon is critical. Avoid having the pursuit of business value turn into simply a lab experiment with new technology—be ready to adjust course and stay committed. The promise of AI is real, but as you explore the data, the findings may surprise you.
I love it. I know you do a lot of work externally as well to help democratize AI and take others along on the data literacy journey.
Yes. That doesn’t mean making everybody a data scientist, but it calls for establishing data literacy throughout McDonald’s. In addition, we want to expose the community to what we’re doing and how we’re doing it. We work with numerous nonprofit organizations to get younger generations interested in this space—not just as a technology or science thing but as something that is pervasive in their lives. As an AI community, let’s pay it forward.
Editor’s note: This is the first in the series “AI From the Front Lines,” which will go beyond the hype to reveal the opportunities and challenges enterprises can realize through AI and data analytics, featuring the real-world experiences of enterprise executives in conversation with Beena Ammanath, executive director of Deloitte’s AI Institute.