Data analytics and AI are the future of internal audit and fraud investigation

Data analytics helps people better understand their business and see weaknesses and inefficiencies more clearly. Analytics can also increase revenue through pricing optimization or analyzing margin or costs to improve the efficiency of the manufacturing process. Many internal audit departments, however, are just getting on board with this trend.

Smart Business spoke with Kirstie Tiernan, data analytics managing director for BDO, Chicago at BDO USA, LLP, about data analytics, especially as it relates to internal audit and why artificial intelligence (AI) adds even more value.

How are employers developing programs around data analytics?

Data analytics requires people, tools, infrastructure and IT, and it’s an involved investment. That’s why it’s best for employers to start developing programs by focusing on one area, such as accounts payable. Once a program analyzes variables like duplicate vendors and payments, employers can expand into accounts receivable, journal entry review, payroll, analyzing customer behavior or pricing optimization. Employers might also benefit from working with an outside vendor, and its tools and subscriptions, so they can evaluate their baseline before purchasing anything.

What are some risks to be aware of?

First, avoid collecting garbage data that requires time and effort to clean. When looking to make better use of data, you should review your information governance policies. How have you collected data? What data are you collecting? How long are you retaining it? Who has access to it? Are you collecting it in a way that you’re able to verify it? It’s critical to have a method to improve the quality, so your analysis is useful in the end. Cybersecurity is a concern as well. Make sure you’re storing data appropriately and that third parties with access to your data are vetted and secure.

One pitfall of data analytics — especially with internal audit — is that people tend to focus on generating reports. They can get overwhelmed and find it hard to wrangle those results into value. As you’re creating your analysis program, make sure you have the results in mind. If you get 5,000 exceptions from an analysis, it’s not a good analysis. You want limited results and as few false positives as possible. That takes upfront planning. The goal is targeted analysis.

The holy grail of analytics is an alert-based program. A restaurant that looks at voided transactions on a monthly basis, for example, might find it more valuable to receive emails flagging where and when voided transactions took place. Those emails can include which voided transactions look fraudulent based on the knowledge of why voided transactions are an issue. This format moves you away from cumbersome report reviewing and toward real-time analysis of specific problems you need to address.

How does AI aid fraud investigations?

Rather than analyzing samples of data, AI incorporates statistical and advanced analytics to review entire populations for anomalies. It’s a new level of analysis that reviews an entire population of data to find transactions that look different. As the advanced algorithms get smarter, looking across more industries and company data examples, they identify anomalies more quickly and efficiently.

If you suspect fraud, targeted data analytics can look for variables like your typical round dollar payments or users with inappropriate access. When you’re unaware of fraud, however, it can be difficult to know where to run specific analytics. This is where AI is invaluable, running millions of data points through algorithms for a quicker focus and narrower scope. Rather than running 50 reports and sampling the results, AI looks at the entire data set. For example, in one investigation, the client had 10 million journal entries over three years of data. It knew it had a fraud issue, but wanted to understand the fraud’s scope and if there were other issues beyond the ones it was aware of. Plus, the client had three days to get back to its auditors. BDO used Mindbridge to examine all 10 million entries. Of the top 10 accounts of highest risk as noted by the tool, two of those accounts had fraud. When you don’t know what you’re looking for or you’re unsure of the scope of the fraud, advanced analytics incorporating AI can provide a quicker and lower-cost application of analytics.

Insights Accounting is brought to you by BDO USA, LLP