Blogs

Use Operational Analytics to Quickly Detect Banking Fraud

In this blog, I’ll share three steps to help drive a faster learning cycle for fraud operations to help prevent and detect banking fraud. This process is highly visual, low tech and requires minimal project investment. The focus is in how to drive improvements for better fraud detection and prevention.

 

As a fraud prevention and detection expert, I receive many inquiries asking, “What is the new fraud detection tool?” Or “Do you know of anything that works well to detect a certain type of fraud?”  And while yes, there are many great tools on the market, fraud often comes back to the basics. The key to successful detection is rapid learning.

As a regulated financial institution, you must worry about change control, compliance reviews, IT requirements, testing and change management. Fraudsters can test anything and implement very quickly. Therefore, you are at a structural disadvantage and must gather more information from how you run your operations.

 

Eliminate Waste to Pinpoint Vulnerability

Getting more from operations is a new concept for many organizations. Growing up in manufacturing I saw the power of managing operations with a zeal for eliminating waste. We defined waste as anything the customer would not pay for. Every unnecessary step, component or extra handle of material was waste. Spotting waste was easy by simply watching the manufacturing line.

However, in our digital call centers we cannot physically see wasteful aspects and we often rely on weekly or monthly reporting that looks back. This adds delays that bad actors exploit.

 

Build Routines to See the Fraud Within Your Digital Environment

Identifying fraud by looking closer at operations takes three steps:

  1. Understand how the work is completed
  2. Gain knowledge of where the control points are
  3. Create routines that drive the virtual to a physical environment

 

Step 1:  How is the Work Completed?

The powerful tool of process mapping helps to understand what comes in, how it’s handled, where it goes and what must be touched again.

If we look at the typical card fraud alert, we can illustrate a basic flow.

Basic Fraud Process Flow Example

Process flow for how to detect fraud

A transaction occurs and creates a fraud alert (Input). That alert is routed to the fraud team to review. The outcome of that review can be “clear the alert no fraud,” “confirm fraud, communicate with the consumer,” or “pend the alert for review later.”  Therefore, the process outputs are either fraud or not fraud. Rework is pending for the future (extra work) or a poor decision (fraud or not fraud).

Converting these flows into pictures provides a clear view of how work is done. It also highlights steps that create waste. By converting the digital world to a visible picture, it becomes clear where the handoffs occur or the existence of complex steps that can reduce the speed of response.

“Converting process flows into pictures provides a clear view of how work is done.”

Step 2: Where Are the Key Control Points?

Once your process maps are in place, you can understand/see where the key control points exist. You’ll most often find these areas where items build up or require longer manual processes. If staff aren’t available to do the manual work, the entire process gets clogged.

In our alert example we can identify some control points such as:

  • Length of time to review an alert?
  • How many consumers respond to the two-way notification of the fraud alert?
  • What is the decision accuracy of agents reviewing alerts?
  • How often do customers call in before you review a fraud alert?

Another control point is key regulatory timers to meet requirements such as REG E and Z actions. Operational analytics do not need to be complex.

Key Control Points Example

Control Points Example for how to detect fraud

These control points tell operations teams what is happening, how fast, how efficient, and what impacts the customer.

 

Step 3:  How to Bring the Virtual into the Real World

Steps one and two rely on data and provide a virtual backdrop. Now, you need to move the virtual into the real world. Routines that engage operations, management and leadership will do that.

You can create a routine by where the work occurs that highlights the basic process, defines the key control points and houses the source information. Many may hear these referred to as GEMBA Boards from LEAN process management.

These boards are nothing more than a meeting spot to provide a tangible location for a routine. The board makes it easy and quick to update. Then, routines are created to review the process. The output identifies if the process as defined today is working. If not, you need to find where it’s struggling and what actions to take.

GEMBA Board Example

GEMBA board showing how to detect fraud

You’ll need to define working as a low waste process. Using our previous example, the routine board might show the following:

  • The alert volume for spyware services has increased 10%.
  • Agents notice that a new merchant is generating most of the volume and the price they are charging varies between $9.89 and $39.99.
  • Additionally, you see some responses to consumers confirming they have authorized the purchase to help stop malware.

Individually these items could go unnoticed. But when you combine the merchant behavior with consumer inputs it is likely that this is a malware scam that entices consumers to buy a product. How fast can your routines and models react to this attack? The visual you’ve created will pinpoint what’s happening and give you a clear picture of the actions to take.

 

Yes, These Steps Do Reduce Exposure and Limit Fraud Impact

Capture an understanding of your operations by leveraging process maps, using basic operational analytics and coupling those with defined routines. I’ve given you a simple sample above. Let me know if you want to see how a more complex Gemba board works.

All three steps outlined above can drive a faster learning cycle for fraud operations. While the cycle time will never meet those of fraud perpetrators it can help to reduce your exposure and limit your consumer impacts.

I have witnessed these routines and saw a response within six hours of the first victim. This will not always be the case. But consider how many consumers you could help and fraud loss you could avoid even if you detect it a week later before the scam ramps fully. These steps provide a low cost and simple to implement process that can improve fraud operations.

 

Get a Clear Picture and Detect Banking Fraud

If you’re ready to look at your operations to help monitor and identify fraud, Bridgeforce can help you. We will talk through your operations and processes, develop process maps and a visual board and provide you with the best steps to take to start your path to rapid learning.

 

Contact us to begin your journey on low tech fraud detection.

Have a question about this article?

ASK Dave Sanders ,