Banking technology specialist Solaris is using Snowflake’s Data Cloud to boost the productivity of its data team and to help end users create their own products.
The company, which provides Banking-as-a-Service products to finance firms around the globe, wanted to make more of its existing data assets. The starting point was having a working definition of an enterprise-wide data strategy, according to Roy Ben David, Group Director of Data and Analytics at Solaris:
First, we defined what kind of objectives we wanted to solve, the business processes we wanted to tackle, the regulatory requirements, and the skills gap we needed to fill. Then we chose the technology that would bring our vision to life.
After analysing the products on the market, Ben David’s team focused on whether Snowflake was a good fit with strong governance and security requirements in the banking sector. The company was also keen to take advantage of the flexibility offered by the Snowflake platform, particularly around costing:
I’m not a fan of spending a lot of money and then doing the proof of concept. I like Snowflake’s consumption-based model, which means they can grow together with us.
Ben David wanted to ensure that moving to new cloud-based technology didn’t simply replicate some of the firm’s existing challenges on a fresh platform. He wanted to do two key things – boost his team’s productivity, and empower business users to build their own data products:
I wanted to make my team more innovation-focused and closer to the business. I wanted us to stop being a bottleneck and to focus on solving business problems. We also wanted to bring the different domain teams that we have in the company closer to their data. Then they would not only have better visibility of high-quality data, but they’d be empowered to own and develop data products themselves.
Ben David believes domain experts should identify data-led opportunities, while a data team like the one he leads should focus on empowering users. Supporting users means creating a trusted source of truth and a new kind of mindset that encourages innovation:
That includes introducing machine learning to the company to build better and more sophisticated products that will bring more advantages against our competitors and not limit ourselves only to classical reporting in analytics.
Embracing a new platform
Moving to Snowflake required a significant amount of effort. With 70 data sources, a complex migration was anticipated. However, in the end, the move to Snowflake took just nine month and completed in January 2023. Ben David recalls:
We identified the different area of the business and classified them as domains. We didn’t only distribute data to these different domains, but we also located data products within these domains. After nine months, the stakeholders within these domains could start to build new data products on top of Snowflake.
Post-implementation, a training and support package from Snowflake has proved crucial. Solaris has used Snowflake Professional Services to optimize the platform at a deeper level as the project has progressed. Ben David explains:
I wanted to think about how we could use Snowflake to work in a more efficient way. They helped us go through all our data pipelines and our data connections, and that worked really well.
Solaris is now using Snowflake on a day-to-day basis to give people across the business access to data in a safe and secure manner. In the past, only between five percent and 10% of Solaris employees had access to data. Today, 40% of the organization has direct access to Snowflake and these professionals use the platform regularly:
It’s not just that our people have everything they need for work, but we can also control access management.
Employees have now started to identify data issues, says Ben David. Crucially, it’s now much easier for end users to create their own data products:
You don’t need to be technical. It can be a basic report, but it can be something more sophisticated. Employees can connect their applications to Snowflake, consume live data directly, and make everything more automated. Now, people are talking about Snowflake as something they use to build better processes.
Exploiting emerging technology
The result of this rapid adoption process is the potential to identify new opportunities. One area of data-led innovation is Machine Learning (ML). By using Snowflake to power regulatory reporting, Solaris staff can build and train ML models in AWS SageMaker that are used in fraud detection and Know Your Customer (KYC) checks:
We identified some business problems. We gave the target to the data scientists to build models that will solve these problems. They could go to our KYC domain on Snowflake, where we centralise all the insights, move it easily to AWS SageMaker, train a model, and get the results they want.
The aim now is to increase the usage of Snowflake. He wants to build additional self-service interfaces for users through Streamlit, which is an open-source framework that allows professionals to build new data apps:
This will be integrated fully into Snowflake, so we don’t overwhelm our stakeholders and can give them added value. We are looking at use cases to start building interfaces that will increase stakeholders’ use of data.
The company is also keen to find ways to use Artificial Intelligence (AI) in line with regulatory requirements across the finance industry, concludes Ben David:
For example, we’re looking at if it can be used to help us read unstructured data that’s coming from PDFs or if it can help boost search capabilities for our non-technical stakeholders. That capability might bring more users to Snowflake, who can ask human questions and the machine will translate it to code.