Our business is based on automating the monitoring and analysis of 100% of the transactions in an organization’s business processes. For many of our customers, this means reviewing transactions in the T&E, purchase card, and procure to pay processes. When we first meet with organizations, some express concern about whether the cure of automated monitoring might be worse than the disease of errors, misuse, abuse, and potential fraud within their transactions. Their concern is that false positives will require them to slog through a large number of transactions that they would not have otherwise had to review. This is a valid concern to explore.
The good news is that our experience indicates that increasing monitoring from reviewing samples to automatically monitoring and analyzing all transactions actually decreases workload while increasing the effectiveness of the work.
Let’s look at the situation for a company that processes 10,000 expense reports per year. We work with companies who prior to implementing Oversight Insights On Demand (IOD) for T&E were manually auditing 20%-100% of their expense reports. For our hypothetical company, let’s assume they are only auditing 5% of their expense reports today. This percentage is the same as the sample size for control testing in most organizations. This means the following would be true:
Audited expense reports – 500
Audited transactions (expense report lines) – 5,000
Valid exceptions (at 1.5% of the total) – 75
False positives – 4,925
Let’s assume that false positives for Insights On Demand are as high as 50%. (Normally we see false positive rates well below 50%) If this same company was automatically monitoring and reviewing 100% of their expense reports with IOD for T&E, the following would be true:
Audited expense reports – 10,000
Audited transactions (expense report lines) – 100,000
Total IOD-identified exceptions (3% of the total) – 3,000
Valid exceptions (1.5% of the total) – 1,500
False positives – 1,500
It seems counter-intuitive, but IOD for T&E actually presents fewer total findings for review by humans, than the total number of transactions reviewed as part of a 5% sample review. And there’s 1/3 the number of false positives.
IOD for T&E automates the process of performing data tests, correlates the results of the tests, and considers the correlated results in the context of all transactions. These correlated results are based on what is “normal” for the individual traveler, all travelers, the company, the geography in which the expenditure was made, the type of expenditure, and the item and vendor involved. Inevitably, there will be findings that for whatever reason are not valid. Perhaps the traveler involved is an exception. Maybe the event that precipitated the travel was an exception. But even these “false positives” have value in providing insight into policy exceptions and the differences among travelers. The bottom line is that less work will result from an effective automated monitoring program than the work involved in a normal sampling activity.