Optimized Fraud Detection Efforts Improve Client Satisfaction using Power BI

Overview

A well-known national banking institution, with over a century of experience and a focus on integrity, sought to measure the impact that its Card Fraud Prevention Program was having on overall client satisfaction and retention. They also wanted to measure the success of this program in relation to the impacts on their clients and respond rapidly to changes in fraudulent behaviors or to rules which may have lost their effectiveness.

The Challenge

The banks experts recognized that fraud prevention efforts can cause clients frustration, as they may experience confusion, embarrassment, and anger when they receive a card decline notification and don't understand why. However, the fraud team was lacking a way in which to measure the real impact that legitimate transactions that were flagged as declined were having on individuals, or on the bank. The fraud data, where the team could find answers to these questions, was siloed away from the main bank systems, which might contain satisfaction surveys or client call information. This disconnect meant that it could take weeks or even months to learn which transactions were truly fraudulent and which were false positives, complicating efforts to keep up with fraud trends and stay nimble in implementing or pausing fraud detection rules.

As a result of the data silo, advanced knowledge of the source data was required to retrieve any insights. Queries could only be run manually, and the source did not have any data dimensions to help interpret the data. A daily view of the data was not feasible, and regular “big picture” check-ups were difficult. When a rather important customer’s legitimate transaction was caught by the fraud rules, they knew it was time to implement Power BI reporting and quantify the impact that the fraud prevention program was having.

The Solution

BlueGranite collaborated with the fraud team to create repeatable queries and transformations to feed Power BI data models designed with modern best practices. Dimension tables were created and codified, with a maintenance page to rapidly highlight dimensional members needing to be updated. The data refreshes could now be automated and run daily, allowing the fraud team to see changes in card usage and fraudulent activity and react quickly to those shifts. Providing these data models also democratized the data, as it no longer required a subject matter expert to prepare a report each time; all team members can view and explore the data to optimize the fraud prevention program.

On the data visualization side, BlueGranite provided overview and alerting views of the data to rapidly draw attention to areas of concern. Utilizing the modern AI features built into Power BI, BlueGranite created decomposition trees, key influencers charts, and anomaly detection to identify behavioral trends, the factors driving declined transactions, and unusual activity that might need to be addressed. On top of that, we created an event study to determine whether card fraud declines caused bank customers to use their card less often or close their account. This causal analysis quantified the impact fraud rules were having on customers and can be used to quantify the impact they have on the bank’s retention and revenue. Finally, we introduced new ways of measuring expected ultimate fraud and fraud development factors, so they could see whether the current reported fraud is in line with expectations.

The Results

Fraud data can now be viewed daily, by any member of the team, via visuals that eliminate the manual analysis and cognitive load required before. AI analysis is now built in, giving them never-before-seen insights into their data. They have a cohesive view of their fraud rule performance and customer impact, and for the first time, can react and make changes without having to wait weeks or months for the data to be available and mature. They are now able to target rules with high customer impact and low fraud prevention for optimization or removal, and they can continue to identify more as rule performance changes over time; the alerting view saves hours of line-by-line rule review. This has an immediate impact on team empowerment, the bank’s bottom line, and on customer satisfaction.

To learn more about how BlueGranite can modernize your data consumption and provide winning business insights, contact us today.