Salim Syed, vice president and head of engineering at Capital One Software, joined Duroseau for a fireside chat, “What It Means to Be a Data-Driven Organization: A Fireside Chat with Under Armour and Capital One.”
Syed described what it means for his company to be data-driven. “Capital One realized the power of data and technology that can create this amazing, individualized experience for our customers,” he said. “And if you think about banking, it’s really about managing large amounts of data securely, safely, accurately.”
Syed said that there has been an explosion of data in the cloud and that companies must adapt to manage it. “What we’ve learned is that you really have to think differently about the capacity now, the scalable back end, the real-time aspect of it. So, think differently about producing data, consuming data and governing data at that scale.”
Finding the Right Platform to Drive Business Outcomes
“The first step was learning as a leadership team and organization that we need to get from a siloed way of doing things to a more structured governance way of doing things that allows us to be more engaged and intuitive,” Duroseau said. He explained that Under Armour’s IT and data teams developed a framework that dictated several decisions relating to their technology stack.
Duroseau said the company prioritized selecting a data platform that allowed it to put all of its data in one place. Ultimately, the choice was Snowflake. “And then, we’re going to wrap that around a really true, solid framework where we have three core pillars — data governance, engagement and flexibility — three core pillars of our framework that are really the foundation of our data strategy,” he said.
Syed said that Capital One had a similar experience. “This was six years ago, and we were looking at moving from on-premises to the cloud,” he said. “For us, it was really about scalability. Every data platform we looked at at that time had some kind of limitations. With Snowflake, their separation of storage architecture broke through that limitation and allowed us to go at the speed of the business.”
When Capital One moved to the cloud, Syed said, the company experienced an explosion of data. “For a centralized model, it becomes very difficult,” he said. “You don’t have domain knowledge at that scale. The centralized team will become a bottleneck to this, to innovation and to providing business value.”
Syed explained that Capital One’s need for scalability caused it to take a different view on its operating model. “We built self-service tooling, where centralized teams could go and enforce their policies and guidance, guardrails, and best practices,” he said.
“And then we handed that over to the back businesses to say, ‘Go ahead. Now you bought the data. You own the processing infrastructure. Just make sure you operate within the tooling.’” That model allowed users to go at the speed of their business “while making sure that centralized teams were not freaking out because they didn’t know who was managing the database,” he said.