Case Study: Scaling real-time analytics with your organization

Shares

When online-based organizations plan for scaling their business, adapting analytics capabilities for that growth might not immediately come to mind. But it should. Innovative companies that want to ensure they are both highly effective and highly efficient need to determine a go-forward growth plan for scaling their analytics along with their company. Let your analytics capabilities drag behind and the competition will only get faster, produce better decisions and results, and provide an electronic-driven world with the speed and convenience they’ve come to expect.

In 2013, Enova International, a technology and analytics-driven online lender, reached this pivotal point — scale analytics or let growth fall to the wayside. Our company’s analytics originally ran on a homegrown system. While it enabled fast and accurate underwriting decisions, we knew we should evaluate new options in order realize efficiencies and prepare our analytics capabilities to scale with our growing business.

We dedicated a year and the time and talents of several members of our analytics, IT and software engineering teams to imagine, design and implement a new real-time analytics engine called Colossus™. In its first year of deployment, Colossus was directly responsible for $7MM in incremental profitability.

Building a new system required buy-in from others across the business. After all, it took a large investment in time, capital and people resources. The main driver that cleared the way for Colossus was actually our company value of Best Answer Wins. Since this value is deeply rooted in our culture, organizational politics served no function when it came to deciding how to scale our analytics. Instead, we focused on building the business case for a more sophisticated decisioning platform. By simulating the impact of better models in production, we were able to show how this platform would not only drive volume and profit, but also provide a better customer experience through faster response times. Even under relatively modest assumptions, it was clear the investment would produce ROI quickly.

Our company value of ‘Best Answer Wins’ is deeply rooted in our culture. Organizational politics served no function when it came to deciding how to scale our analytics. 

In building Colossus, we wanted to address a few common issues: model deployment speed that is not optimal, a tedious reconciliation process, and frequent required re-engineering of the analytics system due to tight integration with front-end applications. In addition to solving these issues, Colossus has had an immense quantitative and qualitative impact on Enova’s products. Average model deployment time has decreased from about 1 month to 1.5 weeks. Faster model deployment produces lift, which results in more direct financial impact. In addition, the platform enables faster decisions; average model response time is a split second. For the customer, speed of decision-making provides a better experience, which leads to higher customer satisfaction and conversion. Finally, Colossus gives Enova the ability to centralize and streamline its fraud defenses, which increases fraud prevention while improving the customer experience through fewer manual verification steps.

From a sheer volume standpoint, Colossus has the capability to handle massive amounts of data. For fraud and underwriting, it currently runs more than 100 algorithms, handles more than 1,000 variables, and accesses 10.5 TB in managed datasets. In addition, the engine has handled more than 100 million unique customer interactions. The system is also integrated in real time with 20+ internal and third-party systems to fetch and parse both structured and unstructured data quickly and efficiently for model scoring and decisioning.

Colossus has had many benefits for our team as well. The platform has allowed our 50+ analytics team members to work in whatever language they’re most comfortable with and build more complex models. It also continually improves the skills of our analytics department by introducing them to a wider variety of model types that they can then build and update quickly. And a happy, motivated team of highly skilled employees is more valuable than any system we could build.

Looking to the future, Colossus provides Enova with a real-time analytics engine that will scale both in volume and complexity as well as both internally and externally. Team members can implement more intricate machine learning algorithms easily into Colossus, add additional nodes, and create more instances of it – which means we can scale without new technology or additional investment.

While building a scalable platform like Colossus may not be a realistic option for every company due to the immense resources in time, talent and money that it takes, the fact remains that effective, scalable analytics capabilities should be a priority for every organization. In an increasingly competitive business environment, even the smallest advantage in analytics can make a huge difference. For Enova it certainly has.

By Joe DeCosmo:

Joseph DeCosmo is the Chief Analytics Officer at Enova International. Joe joined Enova as Chief Analytics Officer in 2014. Prior to working at Enova, he served as Director and Practice Leader of Advanced Analytics for Chicago-based West Monroe Partners. He also held a number of executive positions at HAVI Global Solutions and the Allant Group. Joe received a BA in Economics from Lewis University and an MA in Economics from the University of Illinois at Chicago. He currently serves on the boards of the Chicago Chapter of the American Statistical Association and the UIC College of Business Administration.

Shares

Related Posts

Leave a Reply

Corinium Global Intelligence is registered in England & Wales, number 08520994. Registered office:
Brook House, School Lane, South Cerney, Cirencester, GL7 5TY.

Share This