Fraud analytics is a somewhat abstract concept to explain, but becomes easier once you break it down into its core concepts – analytics, the analysis of data or statistics, and fraud – wrongful deception intended to result in financial gain.
Taken together, fraud analytics can be described as the analysis of data to detect financial deception. These procedures can be as simple as scanning a disbursements ledger for proper approval, to building a neural network predictive model to detect fraudulent credit card activity. Fraud analytics has evolved over the years into a very complex field that consists of combining business knowledge with statistics and computer science to better identify and understand the actions and intentions of fraudsters, and even to stop fraudulent activity before it occurs.
Descriptive and Diagnostic Fraud Analytics
Two of the main types of analytics often performed are:
- descriptive analytics – answering the “What is happening?” question.
- diagnostic analytics – answering the “Why did something happen?” question.
In practice, descriptive analytics often involve the use of business intelligence tools to summarize and display data to business leaders in the form of dashboards or other types of reports. For example, metrics related to employee spending, such as the number of expenditures at different spending levels, could be gathered in a dashboard for the examinations of trends. Once this is done, trends are often identified, such as a higher proportion of expenditures at a given threshold.
Diagnostic analytics would then determine why this trend is happening. For example, the higher proportion could be a result of receipts being required for all purchases over a given amount.
Predictive and Prescriptive Fraud Analytics
Once the “What is happening?” and “Why is it happening?” questions have been answered, it is often useful to take fraud analytics a step further and answer the questions of “What’s likely to happen?” and “What to do next?” These questions relate to predictive and prescriptive analytics, respectively.
A predictive fraud analytic related to the previously described employee expenditure scenario would be to determine the likelihood of an employee to commit fraud based on past fraudulent events. This would take into account characteristics of both the expenditure, such as those just under the amount requiring a receipt, and the employee, such as their role, duties, or even yearly salary, to come up with a “profile” of employees and expenditures most likely to indicate that a fraudulent event will occur.
Prescriptive analytics can then answer the question of what to do next. For example, if it is determined that employees under a given level of salary are more likely to commit fraud than those with higher salaries, a higher level of review could be required for these employees’ expenses. Prescriptive analytics would also evaluate the cost and benefit of the solution to make sure that the organization isn’t spending more trying to catch the fraud than it would be losing from the potential fraud.
Overall, fraud analytics is a fascinating domain that can bring enormous value to organizations. With a proper plan and infrastructure in place, not only can organizations identify fraud that is happening within their organization, they can proactively implement policies that would stop fraud before if even starts.
If you have any questions regarding potential fraud schemes, internal controls, or the use of data analytics to detect fraud at your organization, please contact Andrew Trettel, CPA, CFE, CVA at 412-697-5436 or Brian Webster, CPA/ABV/CFF, CVA, CFE, CMA at 412-697-5307 of the Business Advisory group at Schneider Downs.
This article is part of a series supporting International Fraud Awareness Week 2020, additional entries are linked below for reference:
- International Fraud Awareness Week 2020
- Fake News or Fake Truck? Hindenburg Alleges Nikola One Videos Were Faked
- Employee Refund Fraud at Amazon
- Coronavirus Fraud: Unfortunately, We Told You So
