How Data Analytics Can Illuminate Fraud Schemes
/
FROM THE RESOURCE GUIDE
Jeremy Clopton, CFE
Organizations that implement proactive data monitoring detect frauds 58% faster and experience losses that are 52% lower than organizations that don’t, according to the ACFE’s 2018 Report to the Nations. As these numbers indicate, harnessing your organization’s data can have a significant impact on your efforts to detect and prevent fraud.
But introducing data analytics into your fraud-fighting toolbox is not a cut-and-dried process. There’s no universal checklist to follow. Every organization has different data to work with, and unique priorities and organizational goals. It takes a hands-on approach to figure out what will work best for you and your organization. Here, I focus on one of the most common schemes detected by data analytics and what the analytics process could look like from beginning to end.
Gather Your Data
One of the great things about using analytics for fraud detection is that you can detect just about any type of scheme as long as you have data available, an understanding of the red flags of the scheme and an understanding of the business. That said, one of the most common schemes I’ve detected in my career has been billing schemes, related to either shell companies or employee-vendor relationships. So, let’s take a closer look at why data analytics are so helpful in revealing this particular scheme.
There is a lot of research on the common red flags of billing schemes and examples of how the frauds were perpetrated. That knowledge, combined with the fact there is always a lot of data available in the cash disbursement process, makes using analytics in this area fairly straightforward. The other advantage to using analytics for the detection of billing schemes is the volume of transactions in the cash disbursement process. Without data analytics it would be very difficult to find the fraudulent transactions in these data sets. Analytics allows examiners the ability to analyze the entire population of cash disbursement transactions (checks, invoices and purchase orders), as well as the vendor and employee files to find unusual relationships.
Clean Up Your Data
Before performing any analytics technique, pay close attention to your data preparation. This is a difficult step for many in the data analysis process and one that is quite important. There are a few common mistakes people make. The first is failing to make a copy of the original data set to work with. Ensuring the original data is secured is an important step for demonstrating chain of custody and providing you a fresh start should the data set being analyzed become corrupted for any reason. Failing to document the steps taken to normalize/cleanse the data is also a common mistake. Even if the normalization process is completed without issue, it’s important to know what steps were taken so you can interpret the results correctly. Another common mistake is not taking the time to understand the data. Though all examiners aren’t expected to be IT professionals, it is important to have an understanding of the system, the data and how it is generated before normalizing it.
Execute the Technique
When identifying a billing scheme, some of the more common traditional analytics techniques include a comparison of employee and vendor information (names, addresses, phone number, tax identification number and bank accounts) to identify possible relationships between employees and vendor. In addition, l look for unusual vendor attributes such as having no physical address, duplicate addresses or acronyms in the vendor name. There are also common tests in the disbursement files themselves, such as identifying checks issued on weekends or holidays or in round thousand-dollar increments, or running a Benford’s Law analysis. If you’re looking for a more advanced technique, use textual analytics to analyze email and identify indications of collusion between an employee and vendor.
Demonstrate Your Results
To effectively communicate the results of your analysis, brush up on your data visualization skills. One of my favorite books on this topic is Data Points, by Nathan Yau. It is a great book on the principles of data visualization. A few quick tips to get you started. Before you start designing anything, you need to determine the message you’re trying to communicate. Also, figure out who your audience is and how they will most effectively receive information. Some audiences will require context and pictures, while others may want data tables and metrics. Though there are many additional principles to consider (chart type, colors, context, narrative and medium of communication), if you can determine the message and the audience before you start designing, you will have a much greater likelihood of success.
Find more information on the ACFE’s upcoming course, Detecting Fraud With Data Analytics Workshop, and learn how you will get hands-on experience using real data sets to apply analysis techniques, identify red flags and interpret the results.