Vandalism Fraud Analytics in Banking and Financial Services

Amalgamation of Vandalism Fraud Analytics and Predictive Analytics provides much stronger impact for Bank and Branch business services

Embracing digital transformation is not a choice anymore with more demand arising in COVID and post COVID scenarios across the world where there is a tremendous push for digital disruption. The entire game of digital transformation is becoming ubiquitous by digitizing bank’s business to suit their customer experience in an ever evolving environment. Shifting operations and demand to the digital channels is becoming “the new normal” and is accelerating. With that, comes the responsibility for banking and financial organizations to provide a seamless and secure experience to their customers. Vandalism or fraud cases impacting the ATM fleet for any banking and financial services organization is an area to look into in order to enable making an informed decision in ensuring availability of business and mitigating the impact on customer experience.

We have looked at the financial impact on ATMs with vandalism and ATMs without vandalism cases and identified a strong delineation between riskiest ATMs versus those which are least likely to be vandalized. Those ranked riskier are costing 2x more than normal.

There are predictive analytics models which provide information regarding which devices will fail in future and how to plan for their availability, plan for their maintenance, plan for resources who manage and handle these services and so on. Amalgamation of an intelligence which is more physical such as vandalism fraud analytics can be coupled with regular predictive analytics to give the impact in a stronger fashion.

Globalization is reflected in security threats:

Criminal activity migrates fast i.e. as soon as police enforcement starts being more effective in a hot threat zone, the wave of crime migrates to a lower protection jurisdiction. While the incidents of card fraud and physical ATM attacks are falling, there is also a 40% increase in explosive gas attacks between 2017 and 2018. Some of the auto protection technologies could be effective but they are expensive and hence there is a need to protect and prevent such vandalism cases up front. Identifying and tracking hot zones for vandalism & fraud allows us to recognize those patterns of migration.

Geographic profiling analysis is inspired from some of the modeling in the fields of criminology and epidemiology to get an idea of hot zones or potential threat areas where criminal activities may occur based on historical data. When we consider and apply this concept for preventative maintenance scenario, by tracking the locations of vandalism or fraud we could define areas, rank ATM fleet by risk exposure and apply risk management on inventory to limit the impact of future ATM crime occurrences. Data Science and AI models utilize geo-location data and historical vandalism cases and are trained to rank ATM by the risk of possible future vandalism in the following month. When performing analysis only on vandalized machines, we have compared the following scenarios to see greater impact.

• ATMs vandalized 1 time versus ATMs vandalized >1 time for a period of 6 months or couple of quarters
• ATMs with top 10 scores versus ATMs with bottom 10 scores (low score indicates higher risk and higher score indicates lower risk)

Both groups indicate that ATMs with the worst average score have been hit by vandalism more frequently and for a higher average cost for any firm.

“Low hit score” indicates “higher risk”. This is clearly evident from the experiments performed and ranking holds accordingly. Experiments indicate that scoring is consistent across the two groups for the entire time frame of data analyzed: “ATMs with any vandalism or fraud” versus “ATMs without any vandalism”.

A threat centric approach instead of a vulnerability approach can also be looked at by analyzing ATM video surveillance system where behavior analysis can be performed to identify gestures that indicate a possible attack with the help of machine learning techniques. Pre-emptive steps can be taken by banking and financial services organization for scenarios where, a criminal who wants to blow up a safe ATM / branch area will not have same calm attitude as compared to a person who simply wants to withdraw cash, get account balance, perform pin change or wish to perform any regular and genuine transactions for that matter.

To sum it up, I strongly believe that research and experiments have indicated a pattern that vandalism fraud analytics by geo location is important to tap potential criminal zones that impacts ATM fleet which can be better managed with help of vandalism analytics leveraging advanced machine learning techniques. This has direct impact on availability and hence customer experience plus the operational cost savings for banking and financial organizations. While Predictive Modeling is an key alerting / warning system, Vandalism fraud analytics can help add insights on how to minimize the risk exposure of ATM fleets and further augment the entire digital transformation landscape.

Disclaimer: The postings here are personal point of views from my experiences, analysis, thoughts, readings from various sources and don’t necessarily represent any firm’s positions, strategies or opinions.