This is Digital, Episode 40: Unpacking UPS' Approach to Making Everyone a Data Scientist

In this episode, Joe Brown, Senior Partner at West Monroe, talks with Stefanie Khan, Director of Business Intelligence and Data Analytics at UPS. They explore how UPS uses data science and AI, optimizes data insights, and balances tech advances to drive innovation and solve strategic challenges. Stefanie discusses how empowering every employee at UPS with data science and fostering a culture of learning strengthens the organization from within.

Q&A

You're involved in foresight and forecasting, you're at the edge of technology, and every group inside UPS likely wants to collaborate with your team. You mentioned wanting to scale this across UPS. How are you finding ways to scale your team's impact?

It's honestly difficult to do so. There's a lot on our ask list, but not as many teams to spread the work. Expanding my team is one of the goals for the rest of the year so we can accelerate execution into 2025. Our role is augmenting what we do today, not replacing it. Other IT functions handle process automation using tools like UiPath. Our value-add is providing net new, incremental capabilities. The exciting focus is experimentation—putting our data to the test, incorporating non-traditional datasets, and experimenting with market and industry intelligence, as well as competitive data. The aim is always to augment our current work.

There's always a question of what data can we gather or try to prove whether this new method will work. How do you all approach experiments and what's the level of proof that you look to gather to say, "Yeah, we should double down"?

We typically have a benchmark, and that benchmark is the business results. Our goal is always to see how much better we can be than those business results. Given my lean team and limited time, we start by incorporating a new data set into our algorithms to see how much it improves accuracy rates during the experimentation phase. We track improvements over time to determine if the data set is worth integrating enterprise-wide.

One effective approach is running pilots and A/B tests, applying experiments to specific regions, or product categories rather than tackling everything at once. This helps us prove or disprove theories and assess their impact on the bottom line. The key is to communicate the results and their significance to stakeholders, as gaining their buy-in and driving adoption is often more challenging than the experimentation itself.

You mentioned that one of your goals is to democratize data science across UPS. When you engage with a team to explore new factors in marginal costing, how do you scale your team's efforts in such engagements?

There are a couple of facets to that. The first is defining our role in the project, which is to augment rather than replace the current analysis work. We aim to provide different theories based on our findings, which is exciting because we don't know what we'll discover yet. As we peel back the layers of data, much like peeling an onion, we often find untouched information due to a lack of skills, ideas, or time. Creating dedicated time and capacity across teams is essential. For scalability, we scope the project by determining where to stop, how much time to spend, and the resources to allocate. Currently, this is open-ended as we don't know the full extent until we get there. We're in the process of uncovering these layers, which is a key part of our work.

We touched upon data science democratization. How is UPS addressing the skillset gap in data science internally and what impact does this have on the company's culture?

Like many companies, we've been strict with spending and couldn't hire 100 data scientists. To address the skillset gap internally, we created the Citizen Data Science Development Program, which we've been working on for almost two years. Last year, we ran a pilot with a highly customized program for finance. The goal is to make it an enterprise-wide program. We use a nomination process where leaders identify high-potential individuals with the aptitude for data science, focusing on those with a mindset of change and innovation. The year-long program includes a bootcamp style training and rotational applied learning, where participants observe real data scientists and then execute their own projects. This year, we've more than doubled the program's size. This initiative nurtures UPS culture by investing in high-potential employees, resulting in energized and motivated talent, and a significant ROI for the business.