How S&P Global trained its workforce to be data-driven
When it comes to being data-driven, S&P Global Market Intelligence is taking that goal to the extreme, training the entire organization to be citizen data scientists.
The effort is part of a company-wide initiative to teach employees skills to incorporate models and data science into everyday workflow. It’s a c-level initiative to become model-driven as an organization as opposed to model-driven individuals.
Information Management spoke with Zachary Brown, lead data scientist at S&P Global, who just went through a 10-week program during which more than 130 employees spanning multiple global divisions participated. The course was designed for S&P Global by Domino Data Lab where employees learn: what it takes to train a model; what impacts a model will have on the business goal; and knowing that it was built by someone, and it’s not just the algorithm calling the shots.
Information Management: You recently blogged about your experiences at S&P Global in advancing data science literacy throughout the enterprise. What did you mean by data science literacy and who were the targets of this new knowledge or understanding?
Zachary Brown: We want a forward-thinking, data science first mindset. All associates should have a basic understanding of how we use data science to solve problems. This empowers them to identify data science problems in their own domains.
IM: What was the project or business need that drove this effort?
Brrown: One key focus area for our team is data science community building. We provided an opportunity for professional development and collaboration.
IM: How was data science literacy viewed as critical to its success?
Brown: Data science has become a critical strategic factor for many businesses. A broader understanding of data science and how it impacts our business contributes to the success of our data science initiatives.
IM: How did the organization map out how the process would be implemented, how was it delivered, and who was put in charge?
Brown: This was largely a grassroots effort. I worked with the team to identify core components of the session, and worked with Domino to provide a unified platform for participants to carry out hands-on exercises. Each week, one or two members of our team would lead a guided review session. These sessions covered the material for the week, coding exercises, and also highlighted relevant internal business use cases.
IM: What were perceived as the key ingredients for success with the effort, and how was success to be measured?
Brown: Our key ingredients for success were the interactive review sessions emphasizing relevant use cases, and Domino which provided an excellent unified platform for the coding exercises.
IM: What were expected to be the greatest challenges?
Brown: I expected that our primary challenge would be attendee engagement, and attendance did drop off. But we saw only a very small decline in session attendance numbers, primarily centered around product release windows.
IM: What role did an outside vendor play in implementing and managing the project, and how were they selected?
Brown: Our team was actively using Domino Data Lab, and the platform seemed to be an ideal fit for this type of initiative. Working with the Domino team, we worked out the logistics of course implementation and administration for the pool of participants.
IM: How did the effort go?
Brown: The effort went really well. Our participants provided overwhelmingly positive feedback on the experience and expressed excitement for future sessions.
IM: What have been the payoffs from the effort?
Brown: This initiative served as a template for several key components of our newly formed Data Science University, which currently in its beta phase.
IM: What were the most significant lessons learned from this program?
Brown: My biggest takeaway from this program is how large the drive is within our company for continued technical education and professional development, and how much opportunity we have as a company to help our employees grow.