A Decision-First Approach to Analytics
Friday, September 7, 2018
Posted by: Gary Hotze
Joe DeCosmo, Chief Analytics Officer, Enova International
As businesses continue to undergo the process of digital transformation, they’re actively seeking solutions for digital decisioning — that is, automating decisions driven by data, algorithms and business rules that were previously made through manual processes. In fact, a Forrester study commissioned by Enova Decisions found that 77% of business leaders believe decision automation is important to executing on digital strategy.
However, before business leaders make an investment in building or buying the technology required for digital decisioning, it’s critical to first think about which decisions to automate. I call this a decision-first approach to analytics. Based on my 20-plus years of experience leading and growing analytics teams and capabilities, I’ve built a framework for leaders who are looking to improve their decisioning processes using technology but who might be wondering, “Which decisions can and should be automated? Which will add the most value? Which can be implemented quickly while causing the least disruption to the business? And how do I get buy-in across my organization?”
1: Determine Your Operational Decision Points
As a first step, take stock of all the operational decision points throughout your business – particularly in your customer lifecycle. Operational decisions are a perfect candidate for digital decisioning because they are high-volume and repeated frequently, so they can improve incrementally over time. Examples include verifying a person’s identity or determining which offer to present to an applicant. Understand how these decisions are made, the processes behind them, and the potential impact improving those decisions could have on the bottom line.
2: Start Smart but Small
Once you’ve identified the operational decision points, start small and pick one as a proving ground. There are a couple of strategies that can work based on your business. If your organization is resistant to change, you might suggest testing automated decisioning on one of your lower-revenue products or with a subset of customers. Alternatively, more innovative organizations might be open to automating a high-impact decision for a subset of customers.
At Enova, we built and implemented a digital decisioning platform capable of analyzing an unlimited amount of data from dozens of sources to make analytics-driven decisions in an instant. While it may have been tempting to immediately “turn the machine on” and let it run for every customer decision point across all of our brands, we took a disciplined approach and first rolled it out with a small subset of low-risk customers with one of our brands.
That brings me to another point: while there are clear winners and clear losers when it comes to decisioning, there is always going to be a gray area with any model-driven process that requires some decisions to be addressed manually. Digital decisioning allows you to limit manual processes to where they are required. For example, at Enova, our decisioning allows us to identify clearly fraudulent and clearly good applications so we can focus our manual efforts on the less clear cases.
3: Complete the Transformation
Once you’ve proven digital decisioning on a small scale, it’s time to commit to it and complete the transformation. Expand your digital decisioning capabilities to include more decision points. From there, continue to test and refine your models. While digital decisioning drives immediate results, it also allows for incremental decision improvement over time. As a result, your business can reap the rewards of digital decisioning for bottom-line impact now and in the future.