How The Principal uses analytics to build value and competitive edge

Register now

Last year, 139-year-old Principal Financial Group decided to expand its roster of C-level executives to include a chief data scientist, and appointed Joseph Byrum to the critical new role.

As CDS, Byrum develops and manages artificial intelligence systems in equity trading for PFG. These critical systems will transform the way Principal, a $620 billion Fortune 500 firm, does business, using data analytics to build value and deliver a measurably better results for its 21 million customers.

Byrum's goal is to build PFG’s reputation as an industry leader in leveraging data as an asset. His team is developing and leading a consolidated enterprise strategy for capturing, analyzing and leveraging data across the organization to drive financial performance, and operational and network efficiencies.

Information Management spoke with Byrum about the evolving role of the data scientist, what skills and traits most contribute to the success of a data scientist, and how emerging technologies such as artificial intelligence and machine learning will impact the profession.

Q. Data science earned the title of top job for the 21st Century a brief few years ago. What is the status of the data scientist role now?

Joseph Byrum: The role of data science continues to expand, for a simple reason. It is easy to show that investing in data science improves the company’s bottom line. Or, at least it better be! It started with the operations research movement following World War II, and the results have spoken for themselves in industry after industry for several decades.

Data science is here to stay, but today’s mathematical models are orders of magnitude more complex. It’s no longer something a single expert can handle. Instead, large teams of analysts might need several years to create a working solution to a business problem that had once seemed unsolvable. As a consequence of the scale involved, the data science role involves management — and not just for those in charge of the data science effort.

A key part of the job is finding and inspiring the leaders within each team who can work together toward that big payoff.

Q. What are your expectations for the demand for data scientists in 2019?

Byrum: Big Data used to be all the rage, and perhaps it was oversold in a way that suggested amassing information was the key to success. It’s much harder than that, which is why businesses are putting an increased focus on finding the people who can translate data into value and action — think of the traditional role of an operations research specialist. Individuals truly skilled in the science of optimizing business processes will always be valued because they bring value directly to the table.

Q. What does your role as CDS encompass in terms of day-to-day duties, and broader strategic mission?

Byrum: Principal’s mission is to give our clients — businesses, institutional investors and individuals — an edge in reaching their financial goals. It’s about long-term success, not riding fads in search of a quick win.

I manage our data science teams in developing advanced analytical tools that will give our clients the edge they need. It means creating a vision for a project, then breaking up each essential step along the way toward that ultimate goal. Each team needs a bite-size question or task to complete before advancing to the next step. It turns out that reducing the problem into those fundamental elements tends to be the toughest part of the job, as it requires having a clear vision of the whole and every part simultaneously.

A typical day might consist of meeting with one team presenting the solution to a question, a conference call with a vendor seeking clarification on a task, and meeting with a second team that’s working in parallel to verify the solution of another team. The goal is to keep the teams moving step-by-step toward completion. It’s a process of monitoring, motivating and validating while, at the same time, thinking ahead to what’s the next step, and the next three steps after that.

Q. What personal traits, skills and experiences do you believe have most helped you to be successful in your role?

Byrum: My path is not a common one. I came to financial services from biotechnology, where I started as a quantitative geneticist. In terms of subject matter, the two fields — growing plants and growing assets — are much more similar than you might imagine. Change a few adjectives, nouns and throw in a few obscure acronyms and you might not know I changed fields! One of the most essential elements that transfers readily from industry to industry is a systematic approach to solving problems.

Sometimes a scientific approach to decisions will tell you that the way things have always been done in the past needs to change. I always ask, “Is there a better way to do this?” The willingness to adapt and never stop learning is critical to success.

Q. How do you believe you are most helping to impact the success and growth of Principal Financial Group in your role as CDS?

Byrum: I brought to the table a new approach to developing cutting-edge analytical tools based on proven data science techniques. Principal’s leadership is excited about what it means for our clients. They took the risk of trying a new approach, and we’re seeing the payoff in best-in-class results.

My role includes a heavy asset management aspect, in which the organization must compete differently — we don’t follow the pack. So by coming to this position with a different perspective, without preconceived notions about the way things should be done, my team is able to take an already world-class financial product and make more investment dreams a reality.

We’re doing what nobody else is doing in our field. It’s an ongoing process, because no matter what, we can always do better.

Q. What do you believe are the top skills, experiences and traits that separate the best data scientists from the rest?

Byrum: Nothing is more important than critical thinking. As the data-crunching software packages grow smarter and smarter — and AI will play a role here — the need for specialists in particular forms of analysis will diminish. What will always remain valuable is the in individual who can evaluate an issue objectively. This person puts every assertion to the test, never accepting what everyone else says at face value. Those are the data scientists who can look at a problem and formulate a plan to solve it.

At the same time, since so much of the work is team-based, if you can’t communicate your ideas, it doesn’t matter how smart you are. You want people who bring a new perspective, who add to the cognitive diversity of the team, because that’s what solves difficult problems. What seems to differentiate one talented individual from another is the ability to simplify and explain highly complex concepts. The candidate who excels at that is the one that will be selected.

Q. There are so many changes happening right now regarding data analytics, data management, big data and artificial intelligence, how will these trends impact the skills and traits needed by data scientists in 2019?

Byrum: The benefit of focusing on general abilities like mathematics, science and statistics is that a skilled problem solver will always outlive the fads and the trends. People can always be trained in domain expertise and proficiency with one particular software tool or another. Critical thinking is the one attribute that you either have, or you don’t. The generalist will have the ability to learn and apply the best practices of 2019 to the problems of 2019 far more easily than the particular expert in the ways of 2018.

Nothing is more illustrative of the need to move away from narrow specialization than the rise of AI. Machines are best at completing narrow, well-defined tasks using a known set of rules. What they’re not good at is coming up with ideas and solutions outside those parameters. That’s why the generalist is going to have the edge in 2019. And in 2029.

Q. How do you believe new, emerging and popular technologies such as artificial intelligence and machine learning will impact your industry, and how will leading organizations best tap those technologies?

Byrum: AI in financial services isn’t like what you see in the movies. It’s not a talking robot that is going to run the company. In fact, it won’t even replace the human analyst. Instead, the realistic use of AI is far more grounded in reality, and far more impressive.

Augmented intelligence systems support the decision analysis of the humans already on the job. What the systems do is take an unprecedented wealth of information about, say, a potential investment and displays it to a human advisor in a readily digestible fashion. That lets the advisor jettison guesses or hunches in favor of a solid foundation of data to back up every recommendation. Better information in, better information out.

Such systems take what humans are good at — creative thinking — and combine it with what machines are good at — taking care of burdensome details. So the AI will, for example, monitor and ensure regulatory compliance. It will read more news than any human could possibly read in order to flag any potentially relevant developments or trends. AI will do everything to improve the financial analyst’s choices, but the analyst is still needed to make that choice.

Q. What are the top challenges you have faced in harnessing disruptive technologies, and how will those challenges look in 2019?

Byrum: Plugging bad data into an equation will give you the wrong result, and this is no less true of AI. So ongoing validation is perhaps the most challenging aspect of AI development to ensure the AI algorithms are accurate and they are being fed reliable data.

It’s always tempting to say you’re done once you have a system that can produce a great result, but that result isn’t worth much if it can’t be trusted. The data must be accurate, and the analysis must be flawless. That means before AI can work on its own, human experts need to spend a lot of energy double-checking the datasets and the algorithms before they can be set free.

In practical terms, this process is so involved at the moment that only the larger players in the marketplace will have the resources to develop such systems. One day there will be off-the-shelf packages developed to meet these needs, but until then it’s great for companies to get involved because it is a way to secure a competitive advantage.

For reprint and licensing requests for this article, click here.
Data science Data Scientist Machine learning Artificial intelligence