Analytics Infrastructure for Rapid Growth in Four Months
Anthony M. JerkovicHead of Risk and Data at Novo
Novo needed to set up an agile data infrastructure to support the rapid growth of its small business digital banking products
The 5X Solution
5X deployed a new state-of-the-art data stack in less than a week; within four months, it had managed the ingestion and modeling of Novo’s data and enabled self-service analytics.
Novo now has tools for marketing attribution, user segmentation, and fraud prevention with the flexibility to add more applications quickly.
A FinTech startup with a builder culture
Winner of The Banker’s 2021 Innovation in Digital Banking Awards, Novo has prided itself on its build-it-ourselves culture. For the first three years after the small business-focused FinTech company was founded in 2016, Novo used home-grown data and software.
“We didn’t want to spend a lot of time and effort setting up an enterprise-level data infrastructure before we had a need for it,” explains Anthony Jerkovic, the company’s head of risk and data.
The problem of exponential growth
The need for new infrastructure came fast when Novo’s largest competitor, Azlo, announced in January 2021 it was being shut down by its owner BBVA. Novo quadrupled its customer base to 100,000 accounts in the first six months of that year.
“We could see if we didn’t do something, we were going to have a massive problem,” Jerkovic says. “We’d been smart about the way we’d scaled until then, but our data infrastructure was simply not equipped to sustain the rate of growth we were experiencing.”
The risks of low-quality data
Issues of data quality and governance were already beginning to cause problems. For example, account opening dates were saved in multiple formats, some based on Greenwich Mean Time and others Eastern Standard Time.
“If you’re not piecing together a stack that is well integrated, there can be multiple points of failure,” Jerkovic says.
Clean data would be essential as Novo built advanced analytics and machine learning models. “In any model, it’s crap in; crap out. So you’ve got to make sure you have data quality from the get-go,” he said.
The 5x Solution
Once Novo decided to rebuild its data infrastructure, it explored many options and ultimately selected 5X to manage the process.
“5X gave us everything we needed – the software stack, the project management, and the engineering resources–at a price that was very competitive compared to a U.S.-based consultancy,” Jerkovic says.
Novo was particularly impressed by 5X’s combination of enterprise experience with startup speed.
“We like to work with other early-stage companies,” Jerkovic explains. “They’ve also got a pedigree: Tarush, the founder, ran a very sophisticated data operation at WeWork.”
On the path to working with 5X, Novo first considered and rejected building a new stack in-house.
“The first thing we always say is let’s just hire someone and build this ourselves,” Jerkovic recalls. “I found the ecosystem of data vendors too vast to vet and manage effectively. Bringing in a company with a proven track record to do this for us significantly reduced our implementation time.”
Compared to 5X, Jerkovic says, most consulting firms were likely to be expensive, slow, and not familiar with how to design an infrastructure that can scale with a rapidly growing startup.
“There are plenty of firms out there that do data engineering for technology companies,” Jerkovic says. “I’m not convinced that most of them understand our specific use case and what it means to have a modern data stack.”
Jerkovic adds, “5X is an agile-oriented company like we are. We both ship things very quickly.”
5X assigned a project manager and full-time engineer to Novo. They developed a plan that would deliver critical analytics in 12 weeks. First, they met with Novo to determine its approach and objectives.
Jerkovic says he was particularly impressed with the 5X approach to selecting tools for the stack. “They had strong recommendations for the best-in-class vendors for each layer in the data stack, and they gave the reason why.”
At the same time, he says, 5X was flexible and willing to implement whatever tool Novo selected. In the end, it decided to continue using Metabase for its analytics rather than upgrade to a more powerful tool immediately.
Once the stack was selected, 5x’s automated platform installed and configured the tools in less than a week. Then the 5x engineer refined Novo’s data model, managed the ingestion of its data into the new infrastructure, and built the first round of analytics.
Novo Data Stack
- Data stack design and vendor selection
- Installation and configuration of the stack
- Analytics for marketing attribution, customer segmentation, and fraud prevention
- Self-service analytics tools
Business Value Fast
5X completed the work smoothly, and on schedule, Jerkovic says. “They delivered a new stack that provides the framework for us to ingest and use all of our disparate data sources in less than six months,” he adds.
In the first phase of the project, 5X delivered data tools to support three main applications:
- Marketing Attribution: Optimizing spending by tracking users from the first impression on a website through account opening, funding, and the first swipe of the debit card.
- User Segmentation: Categorizing users by industry, size, and debit card spending then associating those segments with predictive variables such as which external software packages they integrate with their Novo accounts.
- Fraud Prevention: Identifying indicators of potential customers who may be trying to open accounts with fake identities or may be likely to try to overdraw their accounts fraudulently.
Making analysts more productive
5X technology has already made Novo analysts more effective. “From the data team standpoint, things are a lot more seamless,” said Prabhat Rastogi, Novo’s data science lead. “Our previous stack had data distributed in multiple locations, so it was difficult to connect data from one database to another, for example, if we wanted to know the financial performance of users based on the marketing channel.”
The new stack, he says, connects all these disparate data streams. “We can now make decisions from analytics based on data from all our sources.”
The analysts, Prabhat adds, will be able to concentrate on more advanced projects as 5X expands Novo’s self-service capacity so business executives can answer questions without assistance.
Ready for the unknown
With the infrastructure 5X has delivered, Jerkovic says Novo is now prepared for rapid change.
“Our next-gen data stack means we don’t have to think through every possible edge case in the data model in advance,” he says. “You can actually have a very complex database and sort out those problems downstream.”
Now the 5X team is shifting to expanding Novo’s data functionality.
“Now we’re at the point where we are using 5X to go and iterate,” Jerkovic says. “Data engineering is very much a marathon. It’s never complete.”
Novo Platform, Inc. (“Novo”) is the powerfully simple banking platform for small businesses. To learn more, visit www.novo.co.
Novo is a fintech and not a bank. Novo acts as a service provider to Middlesex Federal Savings, F.A., and the deposit and banking products obtained through the Novo platform are provided by Middlesex Federal Savings, F.A.; Member FDIC.
Schedule a call to get data-driven
“We can now make decisions from analytics based on data from all our sources. That's a game changer for a company like us.”
Head of Risk and Data at Bank Novo