Here's a little perspective I thought I'd share for the next time you're scrambling to put dinner on the table: McDonald's serves up meals to almost 75 million customers, or 1 percent of the world's population, every day. Talk about a recipe for potential disaster.
A McDonald's Quarter Pounder with cheese.
Finessing the movement of that many Egg McMuffins and Chicken McNuggets 24/7 around the globe takes an operation that's as smooth as a McCafé vanilla shake. To keep restaurant performance at its best, McDonald's looks for help from Rainer Dronzek, director of operations research, and his 12-member team of data and analytics specialists.
It's a big order for a small team, Dronzek told attendees at the Predictive Analytics Innovation Summit held last week in Chicago. But the more sophisticated its use of analytics gets, the easier that is to do.
McDonald's use of analytics is pervasive, since it relies on its data to glean consumer and business insights, measure restaurant performance, and make decisions regarding equipment, location, human resources, and the supply chain. Dronzek's operations group is just one of many working to keep the shine on the Golden Arches via analytics. "We're all using similar tools and using similar if not the same data, but from different perspectives," he said.
His operations team focuses on how to optimize the equipment, processes, layout, and staffing of restaurants so each can reach its best level of performance. It uses point-of-sale data fed into a global data warehouse from the 34,000-plus McDonald's restaurants worldwide, plus what Dronzek calls "conditions data," or data that's unique to each restaurant.
"Conditions data is the restaurant-level data that defines the unique characteristics of each location that affect performance," he said. These are things like customer demand, arrival patterns, in-store and drive-through configurations, product mix, staffing, layout, menu, and weather. Whereas a McDonald's in Chicago might need just one sweet tea dispenser, for example, one in Charleston, S.C., might need six, he said.
Dronzek and his team work out of the McDonald's Innovation Center, which has a time horizon that's 10 years out. "We look at ourselves as a factory for inspiration," he said.
A lot of what the team does today goes into simulation models that restaurant owners and operators can access on the corporate intranet for a visual depiction of how their stores are performing. "They can put their unique store conditions into the model and start playing what-if? 'What if I changed the number of crew?' The objective is trying to tell the story about what's possible and change the behaviors of people running the restaurant," Dronzek said.
As part of its simulation modeling, McDonald's is using a variety of techniques and technologies, including:
- Quantitative video ethnography to learn the behavior of people using its drive-through lanes
- Eye tracking to study how customers move through a restaurant. The point is to ask and answer questions, such as What are their paths through the stores? What are their interactions with the order-takers? After they place orders, what are they doing? Are they looking into the kitchen or at the menu board?
- Video analytics to track time spent in the store and drive-through, for use in process and conditions measurement
"What's the potential of a restaurant? That's the Holy Grail," Dronzek said. But that's tough, of course, he added. "How can we quantify intent? … We're working on some thinking there."
Outside of the simulation modeling, Dronzek said his team uses a variety of other predictive tools. For example, it can deliver store performance details down to the 15-minute level, so it's able to determine what's working best at a given time. And the team can tell restaurant owners, "If you do this in the future, here's what you'd capture."
It's also working on staffing guides for each restaurant -- a change from four years ago, when McDonald's had a single guide for all. Now, using historical data and other insights, the team can tell a restaurant manager exactly where staff should be positioned throughout the day, Dronzek said.
All of this helps keep McDonald's delivering on its self-description as 'fast, accurate, and friendly,' Dronzek said. "That's measured in the models."
I don't eat at McDonald's too often, but next time I do, I know I'll be thinking of all the data-driven insight that's gone into getting me that hot and tasty Quarter Pounder without a wait. If I do have to sit in a longer-than-desired drive-through lane, I'll be wondering what went wrong with the optimization. Am I nuts, or do the analytics behind your favorite businesses -- fast food or not -- come to your mind, too?
— Beth Schultz, Editor in Chief, AllAnalytics.com