Enova Decisions was launched in January of 2016 as an outgrowth of Enova International’s existing technology and analytics capabilities, which are used to offer online consumer and small business loans to 11 brands in six countries including NetCredit and Headway Capital. Launched in 2004 as an online lender, Enova does all its own analytics for credit risk, fraud, operations and marketing and has 1,100 employees and nearly 5M customers around the world. Enova’s core business relies on easy application, rapid online underwriting and multi-channel service. This requires real-time decisions based on analytics around risk, fraud, marketing etc. To deliver this, Enova developed its own platform for deploying analytics and wrapping these analytics with rules for decisioning. The original platform became limiting so the Colossus platform was developed both to support the internal brands and for sale to their clients through the Enova Decisions analytics-as-a-service brand.
The Colossus platform separated out analytics in a service oriented architecture. Colossus can run a wide variety of algorithms from regression to machine learning. It deploys models built in SAS, R, Python and is integrated with a wide range of third-party data providers. This platform supports all the Enova business and is the basis for Enova Decisions.
Enova Decisions then is a real-time analytics platform for real-time predictive analytics and on-demand decision-making capabilities. Enova Decisions supports both a decisioning interface and a reporting interface. The decisioning API allows requests for decisions to be made that are then processed using rules in Enova Decisions and analytics processed using the Colossus analytic platform. Data and distributions about decisions, scores and outcomes are stored and available through a performance dashboard on the reporting side. Thresholds can be set to trigger alerts and model updates etc.
Enova Decisions is positioned as analytics as a service. Initially Enova Decisions is focused on customer experience across customer acquisition, fraud and alerting and customer operations/growth (retention, debiting and collections). These are the areas where Enova has expertise — Enova has a 50+ person analytics team with experience in these areas — but the platform itself is completely agnostic and can deploy and integrate models developed by customers too. Because the platform was built to support the Enova business, it supports complex decision-making with multiple models, A/B or champion/challenger testing and many rules as a single API call.
The platform runs on AWS and is typically set up and integrated before being billed based on usage. Enova Decisions does an initial set up engagement around integration, third party data integration and data analysis as well as (generally) an initial set of analytic models. Once it is set up, Enova Decisions provide operations support and may develop additional or updated analytics. Enova Decisions tries to keep the up-front integration and analytics development costs low so that customers can focus on the ongoing service cost.
The platform is evolving to support PMML to allow models to be integrated from a wider array of analytic tools, and business user rule management capabilities are under development to allow customers to manage rules directly rather than by working with the Enova Decisions team.
You can get more information on the Enova Decisions website, and they will be included in our Decision Management Systems Platform Technology Report.
