We all know the best decisions are data driven. So why are so few leaders making good use of their analytics teams?
In this edition of Author Talks, a series from McKinsey Global Publishing, Ron Nurwisah chats with Howard Friedman — a data scientist, health economist, and adjunct professor of health policy and management at Columbia University — about his new book, Winning with Data Science: A Handbook for Business Leaders (Columbia Business School Publishing, January 2024), coauthored by Akshay Swaminathan. The book explores data science and its business applications, emphasizing the need for improved communication, understanding of basic concepts, and ethical considerations during data science project management. An edited version of the conversation follows.
Why did you write this book?
It really comes down to seeing a need. I’m very frustrated when I see business leaders who don’t know how to use data science teams well. They make inefficient use of the teams.
Disconnects in communication are also problematic. The team is doing a great job, but the businessperson is frustrated. I want to bridge that gap. I want something that really enables the directors, the VPs, the C-suite to be able to have that conversation, to get as much value as possible out of their data teams. And I want something that allows data teams to feel really good about themselves at the same time, so they know that they’re delivering a great product and everyone’s appreciating them.
Was there anything that surprised you in the research or writing of the book?
The truth is I’ve written many books before. This wasn’t my first or second. But what was really interesting for me was the collaboration with a young, very talented data scientist: Akshay Swaminathan. Akshay and I have known each other for a number of years.
As we brainstormed about how to make this an interesting book, he was the one who generated the idea of using a narrative approach. That’s really what makes this book very unique, really reader-friendly, and more accessible than most other data science books out there for businesspeople.
Why did you choose these two career scenarios as your narrative approach?
We have fictitious characters. They’re working their way through the corporate world in different leadership roles and interacting with data science teams. Steve — by the way, my middle name is Steve — is working for a consumer finance company.
He’s undergoing different rotations and is being challenged to learn different things. My very first experience after finishing my PhD was working for a consumer finance company, where I worked in four different departments in the first five years.
Steve is pretty close to my life, and fleshing out him and his backstory was a lot of fun. We really tried to make the book like a novel where you think of the characters. Kamala, named in honor of Akshay’s grandmother, is working her way through the health insurance leadership space.
Akshay himself is a director in the healthcare data analytics space. Kamala’s story reflects some of his experiences and my work with pharmaceutical companies over the years. The challenges really reflect our own lives and the work that we’ve done as professionals. But it was a lot of fun to mix a little bit of our own personal history into these fictitious characters.
How do we make data science less ‘scary and arcane’ for those who aren’t data scientists?
How do we dispel this fear of data science and the specialization of it? It’s about having a conversation; it’s about communication. The first area where people have to connect is on the definitions so that we agree on what we’re talking about.
In the book, without looking like a dictionary, we weave in the important concepts that business leaders should know about data science so that they can have that conversation. We bring in data science definitions. But then we introduce frameworks.
Whether it’s a data science project or any other type of project, the concepts of project management are the same. How we judge success becomes the same conversation. So we bring in those frameworks incrementally throughout the book and really help people understand how to connect the dots. In the end, data science is a specialized field. But there are lots of specialized fields out there. The more we can bridge the dialogue, the more value everyone gets out of their hard work.
What are some of the essential principles to keep in mind when you’re engaging with generative AI and data science?
Generative AI [gen AI] and some of the incredible capabilities we’ve seen in the past few years are really mind-blowing. An important consideration that I advise for all business leaders is to think about the problem you are solving. Don’t get so excited about the latest and greatest tool. Instead, think about the issue you’re trying to address.
What do you want to make a success of? Are you trying to look at a problem from a revenue or cost point of view? Approach it systematically. If it turns out that gen AI provides you a game-changing solution, great.
Don’t start with the answer. Start with the problems first, then see what the answers are. Often, the best answers start with some basic analytics and then bring in advanced modeling or the latest tools that are out there.
How do we keep ethical concerns front of mind when working with data and AI?
Ethics within data science and AI comes up all the time and in so many different aspects. One of the basic ideas is about privacy, about understanding that people’s personal information could be exposed. And it’s everyone’s responsibility to protect that privacy.
The other items that come up are about the use of data. How is it going to be applied? Is it going to potentially put people at risk? Or is it [the data] going to be applied in a way that is ethical and can improve people’s lives? Assessing algorithms to make sure that they’re not repeating and emphasizing bias is absolutely critical to do.
Is it [the data] going to be applied in a way that is ethical and can improve people’s lives? Assessing algorithms to make sure that they’re not repeating and emphasizing bias is absolutely critical.
There are methods for conducting this assessment. From a business leadership point of view, arming the data science team with important questions is the most prevalent one. Giving them a short list of questions to ask about how they are protecting data privacy is a start.
How are they assessing the potential biases in the algorithms? How confident are they? What are the potential risks? Nobody wants to have that moment when they appear as a negative story on the front page of a newspaper. At the same time, it’s not just about the optics of your business. It’s truly about doing the right thing for your customers, for your employees, and making sure that the data is being used well.
What’s the one thing you want an interested stakeholder to take away from this book?
I really want people to realize that data science is amenable to everyone. It’s accessible. And to be a good customer of data science, you need to feel comfortable with basic concepts and to be armed with good questions and good frameworks.
Once you have that information, go out there, engage with the teams, and really see that you can get a tremendous amount of value by, again, treating data science as you would any other project. That’s the key.
What should a data scientist take away from this book?
I want data scientists to recognize that they have a responsibility to communicate well, to get away from the technical jargon, and to make sure that the customer understands the capabilities and which solutions map to which problems. They also have a responsibility to listen, rather than get so focused on the solutions first.
Data scientists have a responsibility to communicate well, to get away from the technical jargon, and to make sure that the customer understands the capabilities and which solutions map to which problems.
That’s why you see a lot of conversation in the dialogue approach of the book: to make sure that people really understand, “What is the project? What is the goal? What are we trying to accomplish?” Because if you establish that, you’ll find that you’ve already gotten yourself halfway to the answer.
On the other hand, if you walk in with that hammer, then you’re always searching for the nail. We really don’t want data scientists to do that. Listening first, engaging with the customers, and understanding are part of the approach.
I came from a very quantitative graduate school background, for example. I studied advanced statistics and engineering. But when I came into the corporate world, it took me a little while to recognize that to be a successful data scientist and analyst, to produce high-quality work, and to lead teams, I needed to understand the great value that project managers bring to the table, the great value that those facilitators and communicators bring.
They’re the ones who helped enable my success. So, I’m forever grateful for those early years when I worked with amazing project managers who taught me to ask those questions. They taught me to listen and then to tailor my solutions to what the customer wanted, not just to what I knew I could build.
Howard Friedman is a chief data scientist, health economist, and adjunct professor of health policy and management at Columbia University. Ron Nurwisah is a senior editor of McKinsey Global Publishing and is based in the New York office. Comments and opinions expressed by interviewees are their own and do not represent or reflect the opinions, policies, or positions of McKinsey & Company or have its endorsement.
Related Posts
