Fraud control at the press of a button

It is one of the most pressing issues for mid-market businesses – how to stay vigilant on employee fraud when resources are limited.

While there is no foolproof method of preventing fraud, certain internal controls are essential such as the physical segregation of key duties, dual authorisations and restricted access to valuable inventories.

For many small to medium businesses, however, best practice internal controls are not always commercially practical to implement and conventional detection techniques like auditing can be limited in their scope.

What is data mining?

For such businesses, data mining can provide a cost-effective and comprehensive solution to detecting employee fraud.

Data mining is essentially the analysis of large volumes of data to detect abnormalities or unusual trends. Using sophisticated software we are able to evaluate data sets at the press of a button, providing assurance over control gaps with a much smaller financial outlay than traditional methods.

Unlike a typical audit where random sampling methods are used, data mining can easily review all staff, creditors and payments made during the period of the analysis.

So rather than choose 5-10 employees and check samples of pay runs for those staff as happens in a traditional audit, with data mining it is possible to map the entire workforce and quickly identify abnormalities such as duplicate payments, large or unusual payments, an excessive number of payments or payments made outside of the expected times of the day/week when the payroll payments are typically made.

Data mining is generally used to focus on control gaps that exist due to limited segregation of duties, which can be established through an initial discussion around the business. The tool is then used to concentrate on those higher areas of risk, providing comfort that a thorough review is being undertaken to detect any issues.

The process is most effective when a complete data set is provided directly out of your business’ system for all payments made (creditor and payroll) along with the underlying source data for all employees and creditors within the system. We are then able to cross-check relationships to ensure all the key abnormalities are being reported, for example phantom employees or creditors.

Experience and critical analysis are essential to getting the best results out of data mining. Used in isolation it can be just another metric.

For small to mid-market businesses, particularly those with a finance team of less than ten people, data mining can provide a practical and comprehensive solution to monitor potential fraudulent activity without involving a huge financial outlay.

Data mining in action

In a recent engagement, William Buck was asked to discretely look at the expenditure of an employee over a given period.  As part of the initial meeting we established that their spouse worked part time in the business helping out with some administrative functions.

By documenting the business’ processes and looking for control gaps we were able to limit our investigations within a targeted scope, which included the spouse.

Running a full analysis of all payments some significant payments made to the employee’s spouse throughout the year were detected.  The investigation concluded that leave balances had been adjusted fraudulently, with over one thousand hours put to their annual leave balance and then paid out over time.

Following the investigation, it was found that collusion between the employee and their spouse had occurred, with a jail sentence being imposed as a result.