June 03, 2021

AI Plays the Leading Role in Outlier Detection


Outlier Detection. While it sounds a bit like the next TV crime drama — it’s actually closer to a reality show. An outlier is an element of a data set that distinctly stands out from the rest of the data (the black sheep of the family). Outlier detection is simply the identification of these rare items, events, or observations. In business, we speak about outlier detection predominately as a way to catch potential errors and identify risks. It’s difficult to do, mainly because businesses generate so much data. Finding the one little thing that’s not like the other can end up being a full-time job – but that was before Artificial Intelligence (AI) entered the scene.

Mistakes Happen. A lot.

You may have heard that Blockbuster turned down offers to buy Netflix back in 2000. What if Blockbuster was missing a zero on one of its projections? How about the Mars orbiter that’s lost in space due to a math error by Lockheed engineers? Or the time when Bank of America overstated its working capital by $4 billion due to an accounting error?

These are extreme examples of course, but routine accounting errors are commonplace. Customer interviews have shown that 10-20% of journal entries typically need adjustment after being posted to the general ledger. These adjustments require extra staff time to find, correct, and document.

In a typical organization, month-end is the time when errors and omissions are sought and caught as reports are printed and journals are scrutinized. When outlier detection is only performed once a month, it jeopardizes the validity of your business data for the rest of the month. If an error was made shortly after the period close, it might not be detected for three or four weeks. Meanwhile, you were making decisions based on bad data.

Catch Them While You Can

AI has made enough news recently that it’s tempting to think there’s nothing it cannot do. While plenty of tasks are better left to human eyes and hands, business data outlier detection is not one of them. AI is, in fact, the ideal outlier detector.

Accounting applications like Sage Intacct are leveraging AI in ground-breaking and innovative ways. Sage is delivering on the concept of the Intelligent Organization by incorporating functionality like Sage Intacct Budgeting and Planning and Outlier Detection for General Ledger. Sage Intacct’s Outlier Detection uses the power of AI to catch potential errors and risks all throughout the month — not just at month end. It reduces the cost associated with catching and correcting errors, and significantly, it enables finance teams to trust their financial information sooner — without having to wait for the close.

Sage Intacct Outlier Detection uses your organization’s historic transaction patterns to review transactions during the approval cycle and catch transactions that don’t match or seem unusual and send alerts about them to approvers, who can then decide what action to take. It does this with the help of artificial intelligence (AI) and machine learning (ML). The tool “learns” over time, getting smarter and helping to continually improve overall data accuracy for your company.

What Can Better Data, Earlier, Do for Your Company?

Access to accurate, real-time data can help you achieve your organizational objectives and deliver your business strategy. Better business data informs decision making, helps you identify opportunities and trends, allows you to provide better products and services, improves internal workflows, and creates additional revenue.

Outlier detection, like that in Sage Intacct, builds confidence in your business data, powers smart decision making, and helps catch anomalies before they skew your vision. It’s one of the most practical and strategic use cases for AI we’ve seen.

Business insights and resources

senior living software

5 Ways to Solve the Integration Challenges Senior Living Communities Face

SaaS Revenue Recognition

Mastering Subscription Accounting: How To Streamline Recurring Billing and Revenue Recognition Tasks

Domo data warehouse

Do We Have to Move Our Data?

Ready to optimize?