There’s a great line in Jurassic Park, when two of the characters are arguing about the scientific achievement of creating genetically engineered dinosaurs and the skeptic says:
“Your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should.”
I think a slight derivation of this same sentiment is highly relevant for banks today:
“Your data scientists are so preoccupied with whether or not they can predict something, they don’t stop to consider what will happen if they are successful.”
The “use of Big Data, AI, Advanced Analytics, Cognitive Computing” was listed as the top trend in banking for 2019, according to research from The Digital Banking Report. And with good reason. Every area of the lifecycle — from marketing and customer retention to collections and compliance —can theoretically benefit from the application of more sophisticated data science tools and techniques to rapidly growing data sets.
And yet, for many business executives I speak with, the benefits of applying advanced analytics to these use cases remain frustratingly theoretical.
Why?
Put simply, banks spend comparatively too much time trying to figure how to predict something (or improve the accuracy of an existing prediction) and not nearly enough time mapping out how their people, processes, and technology will need to change if they are successful.
I saw this up close during the Great Recession, when the analytics question de jour was “Can we predict strategic defaults?” Turns out, with the right data, you could predict strategic defaults (in which a consumer who is underwater on their home will choose to default on their mortgage in order to keep paying their credit cards, auto loans, etc.) with a startling degree of accuracy.
The trouble was, once they had a strategic default predictive model in hand, most lenders found that it didn’t do them much good. They could identify a specific borrower, living on a block of homes that were all underwater, who was strategically defaulting on their mortgage. However, they couldn’t offer that borrower a restructured payment plan without triggering a ripple of additional strategic defaults down the rest of the block.
Lenders wasted time and money creating strategic default models, not because the models themselves were inaccurate, but because the lenders failed to anticipate how the attractiveness of a loan restructure offer would influence customers who hadn’t yet defaulted. In other words, the analytics didn’t fail. The implementation strategy did.
Here’s How to Do It
Now that we have our cautionary tale out of the way, let’s take a look at how it should be done. Here are three examples of where banks are succeeding in aligning advanced analytics with operational realities in order to drive a substantial return on their investments.
- Anti-money laundering. Today, the vast majority of suspicious activity reports (SARs) are generated by transaction monitoring through scenario-based rules. With the use of machine learning and predictive analytics, FICO could generate 90+% of the same SARs automatically, without any human assistance (not to mention capturing a multitude of new money laundering events missed by current transaction monitoring). This advance in AML detection can deliver huge benefits to banks when it is paired with a sound implementation strategy that ensures regulators are comfortable and that existing compliance officers are reassigned to focus on higher-value alerts.
- Collections. Advanced analytics can be used to segment, at an extremely granular level, delinquent borrowers into different buckets. If a bank can identify a segment of past-due accounts that will cure on their own or with a light reminder through an automated communication channel, then they can improve their operational efficiency by not assigning those accounts to human collection agents. Again, a sound implementation strategy is critical. More accurate collection segmentation (driven by analytics) can deliver huge benefits when banks are prepared to refocus their human collection agents on higher-priority accounts where their empathy and negotiation skills will make the biggest difference.
- Attrition. FICO works with a large telecommunications provider that struggles, like every telco, with churn due to customer attrition. To combat this, they have assembled a “tiger team” of customer service specialists who are remarkably effective at convincing customers not to leave. This Attrition Intervention Group is highly effective, as long as they are focused on the right customers. That’s where analytics come in. Investments in more accurate attrition prediction models, even when the lift in prediction accuracy is relatively small, pay off handsomely because those improved predictions convert into “saves” for the business at an impressively high rate, thanks to the Attrition Intervention Group.
The role of analytics in business transformation tends to get lionized (we at FICO are certainly guilty of this). The truth is that analytics is an extremely powerful tool, but the effectiveness of any tool depends to a great degree on how it is used. Bankers making analytic investment decisions would be well-served to remember this.
