Wharton@Work

September 2024 | 

Data-Driven Success: McDonald’s Analytics with Wharton

Data-Driven Success: McDonald’s Analytics with Wharton

Why would a multinational corporation with state-of-the-art analytics capabilities assign a project to an external analytics team? McKinsey identified one reason in its study of analytics-driven organizations: top analytics performers typically include “an ecosystem of partners that enables access to data and technology and fosters the co-development of analytics capabilities, as well as the breadth and depth of talent required for a robust program of AA [advanced analytics].”1

“We had a few years of data about our social posts,” says Jola Oliver, U.S. director of digital, menu, foundational and behavioral science insights at McDonald’s, “but we weren’t looking at it or modeling it because our social data was always combined with our PR data in our marketing mix models. Our teams have a hundred other things to do, and I'm not able to use their time or their talent to be able to do every kind of modeling for me. We never teased apart the actual impacts of the posts and the engagement on them, looking at whether there are any similarities [or] signals that say things are working.”

After attending Wharton Executive Education’s Analytics for Strategic Growth: AI, Smart Data, and Customer Insights program, Oliver discovered that the academic director, Raghu Iyengar, was pairing teams of students at the MBA and undergraduate levels with organizations seeking analytics assistance. She engaged the student teams to work on a couple of key projects, including the one on McDonald’s U.S. social data. “We wanted to understand whether social conversations had any impact on sales and what types of things would cause our sales numbers to jump. That project turned out to be an amazing collaboration.”

Oliver explains that Iyengar’s team looked at unstructured data — text from posts, comments, likes, shares, reposts, hashtags, emojis, and even multimedia content like images and videos. Unlike structured data, which fits neatly into databases, unstructured data is messy and diverse, making it more challenging to analyze. “This was the first time we actually looked at our social data in totality, and the students were able to glean some cool insights,” says Oliver. “For example, we learned that the value and impact of a social post on average lasts longer than one day, and in some cases it could last up to four days. That means that if you have a negative conversation, its effects could last for a long while, which is why it's important to push out conversations that are really, really pithy and on strategy. Those kinds of posts are much less likely to spark negative conversations. The negative stuff was measured by the Wharton team as much more negatively impactful than the positive ones were positive to our brand.”

“You’re always thinking as you plan your strategy, ‘What can I post next that's going to build my brand?’ But if you're not aware that something negative can hurt your brand a lot more than a positive post can improve it, you might not be as careful as you should be,” Oliver says. Some of the other insights from the project include the fact that short posts perform better than long posts. “It's hard to change an entire strategy, but we did change the length of posts based on the students’ findings. We also learned that posts with a specific call to action are much less likely to inspire engagement and reduce comments from detractors that can turn into long, negative conversations.”

Enhancing the Work of Internal Teams

Oliver says McDonald’s marketing teams and data analysts are “fully able to do work like this. It's all about priorities, and these were not prioritized questions. That’s why I reached out to Wharton. I knew I would get teams of eight to 12 data scientists, and that I would be able to direct and collaborate with them. I knew the data, but they knew how to model it. That's hard to do in eight weeks — it takes about one year to build a functional text analytics model that’s 80 percent accurate. So for them to be able to [get] as far as they did, delivering fantastic results in just eight weeks, for me was very powerful.”

The results were so powerful, in fact, that Oliver tapped Iyengar’s students the following year to study McDonald’s online conversations about the launch of their loyalty program. They wanted to glean new insights about the link between customer service and how it drives, or doesn’t drive, loyalty. Once again, the student teams delivered. “They got coaching from faculty and mentoring from other business partnerships, which meant the sky was the limit in terms of what they could come up with. Because they had so many tools at their disposal, they could try different ways to attack the problem. Seeing that kind of effort was one of the best parts of the partnership.”

But the gains were experienced on both sides. “We taught the students about our data and how we model,” says Oliver. “But we also learned about some of the ways that they're modeling the data and what we can  do in-house differently next time so we can model it better. There was a lot of back and forth in terms of, ‘Here's the math of it, but what's the art in terms of telling the story and having implications for how we do our business moving forward?’”

Professor Iyengar says the collaboration, and others like it, reflect the wisdom of Penn’s founder, Benjamin Franklin: “Tell me and I forget, teach me and I may remember, involve me and I learn.” “These engagements serve as a wonderful example of what we believe at Wharton — students learn best by getting involved and doing,” Iyengar says. “They begin to appreciate that a business question is never answered in a vacuum and that multiple stakeholders are involved and have different points of view. In addition, they also see that what makes any recommendations more actionable is the close collaboration with experts who have domain knowledge. This experience also gives students a significant advantage while recruiting. I feel proud that we are getting students prepared to be the next generation of business leaders.”

For Oliver, though, the practicality of working with Iyengar’s students is the primary focus. “These kinds of partnerships are important because we can't do all the work ourselves,” she continues. “We'll never be able to do it all and keep up with every new method. Having a relationship with an academic institution at the forefront of AI and analytics is important to us because Wharton has some of the smartest and the brightest minds in the business. We can rely on the relationship over time, which is a more robust approach than relying only on your internal capabilities.”

1. Building an effective analytics organization | McKinsey