While many finance firms are still constrained by legacy systems, Capital One has spent the past decade embracing data-led digital transformation, pushing systems to the cloud.
The success of this shift has been such that the bank recently launched Capital One Software, a new B2B software business focused on providing cloud and data management solutions to other companies.
Now the financial services giant intends to use its experiences to help other businesses on the same journey.
According to Salim Syed, Vice President of Slingshot Engineering at Capital One Software, Capital One’s commitment to technology is best exemplified by its cloud-first strategy, which has resulted in the bank closing its internal data centers. As part of that digital transformation process, the bank built its own tools to help it manage the shift to the cloud, he says:
Being first to the cloud, you get to realise not everything's already built for you – the things you need to manage the scale, the cost and the governance. So, we ended up building all these things internally. And these are battle-tested – they are tools that meet our regulatory standards, as well as the scale that's required. And we realised later on that this is something that other companies are going to need, too.
The bank’s first product is Capital One Slingshot, a data management solution for customers of Snowflake, which aims to makes it easier for other businesses outside the bank to adopt the platform. Syed says Slingshot does three key things:
It allows you to automate your Snowflake warehouse, it manages your cost, and it automates your governance. We believe in the federation of the management of data infrastructure. To do that, you need to provide some self-service, easy-to-use tools to the lines of businesses to manage their data platform. And so that's what we did. This approach allowed us to scale at the speed of the business instead of everything coming to the centralised team and creating bottlenecks for innovation.
Capital One doesn’t expect the exploitation of its technologies to be confined to the finance sector. The aim is to help line-of-business managers in any industry to develop a proactive approach to on-demand technology. Syed says the bank recognized early on after adopting Snowflake that a simple dashboard doesn’t provide enough detail on resource use and it wants other companies to take a similar approach:
Not everyone comes to the dashboard. So, then we started thinking about alerting. We wanted to give alerts in as near to real time as possible. We wanted to give a recommendation on how to remediate the issue, because we're assuming the user is not a technical person. We're going to educate you at the same time as we give you a recommendation. And that's what Slingshot does today.
Making the most of Snowflake
Capital One first started talking with Snowflake about its technology in 2016, when the bank was an early adopter of the platform. The firm looked at a range of cloud databases at the time, but was impressed by Snowflake’s fine-grained access controls and its focus on collaboration and security. Most of all, says Syed, the bank was impressed with the scalability of the platform:
Data is like oil to us and we want to empower our analysts to do whatever they need to in order to get insight. So, we allow our people to pull whatever data they need into Snowflake, to do their analytics and get that insight out as quickly as possible. And Snowflake does that; we can just scale to whatever the demand is.
Snowflake technology now plays a crucial role in Capital One’s data management processes, enabling the bank’s 6,000-plus analysts to run as many as four million queries every day. Capital One has built tools in Snowflake that have saved 50,000 hours of manual effort, reduced spend by 27% and reduced cost per query by 43%. Syed says the benefits of Snowflake for the bank are clear:
We really loved the separation of storage and compute. Because for the first time, Snowflake allowed us to scale storage when we needed it and shut it down when we were not using it. We just couldn't do that before. Now, from real-time data intelligence to fraud detection, all kinds of our data workloads are there. Pretty much every single line of business in Capital One is using Snowflake right now.
Syed advises other digital leaders who are thinking about implementing Snowflake technology to undertake a rationalization exercise. Before migrating data from legacy platforms to the cloud, think about what data actually needs to move. He estimates as much as 70% of data might not actually be useful, so think carefully before you shift:
Migrating is not cheap. Snowflake is great, it's powerful, but with great power comes great responsibility – and that's why I think that, before you go big, think about governance and controls that you want to put in. Specifically, if any feature has costs associated with it, don't just release it to your users. Build some controls around it – when to use it, when not to use it. Put the guardrails in and then release it.
Capital One Software continues to think about how to take advantage of new features. The organization is using Snowflake’s recently announced Native Application Framework to build and deploy applications natively in the Data Cloud, which aims to create an integrated and cross-cloud data platform to eliminate technical and institutional silos. Syed concludes:
Whenever I see a cool new feature, it's a building block. I can make something around it. I look at it from the lens of experiences – what kind of self-service experience can my team build so that non-technical people can make the most of data.