The Truth Behind Detecting Fraud Using Data Analytics

In our recent webinar, ‘The Truth Behind Detecting Fraud Using Data Analytics Tools’, data consultant Sunder Gee dispels some of the myths around using data analytics to detect fraud. As he debunks these myths, Sunder reminds us of the important role data analytics plays in audit and shared best practices for preparing data for audits that can immediately be applied to your processes.

Myth: Detecting fraud using data analytics takes just a click of the mouse—FALSE


With the advances in data analytics software, it is certainly easier to do more. But data analytics cannot directly detect fraud. In most cases data analytics is used to determine anomalies, but it is only after investigation and verification that an audit can assess whether a particular transaction is fraudulent. If fraud is detected it is at the end of a non-trivial multi-step data analytic process.


Myth: Data analytics saves time—TRUE


Most auditors, however, reinvest that time into expanding the audit. Sunder explained that every transaction can now be involved or touched against a selected criteria by the auditor using data analytics. Furthermore, auditors have caught on to other ways to apply data analytics, such as finding ways to fix the weakness in internal control systems—which is as important as identifying fraud itself. Over time, control systems may eventually develop weaknesses. With data analytics, auditors can test controls against established parameters across systems from different applications and data sources.


Myth: Data Analytics tools can take care of the whole audit process—FALSE


Human intervention is necessary to investigate suspicious activity. Auditors with an understanding of the business, its practices and procedures will be able to explain most anomalies that appear. Experienced auditors are also needed to note poor system designs that can lead to the control problem. Detecting fraud begins with detecting anomalies and then focusing on high-risk anomalies. The next step is to investigate by looking at documentation, people and procedures. It’s only after these steps are completed that fraud can be verified with analytic tools. Each of these steps involve human intervention to execute, examine and assess.


The reality is that there is no magic wand to finding fraud. Detecting fraud is part of a process that starts even before you begin to analyze data; it requires processes, leveraging the analytic functions available in most tools and, of course, human intervention. More often than not, there will be a high number of anomalies from the data analytics process, with very few actually being errors and even fewer being actual fraudulent transactions. But with professional judgment and some intuition (based on experience with an organization), data analytics will make the process faster, more focused and easier.


Interested in learning how data analytics can be used to detect fraud, flip through our SlideShare deck about using Benford’s Law for fraud detection and audit:



About Anu Sood:

Anu Sood is the Director of Product and Corporate Marketing at CaseWare Analytics and is responsible for the company’s global marketing strategy. Prior to CaseWare Analytics, Anu worked in various roles in the high-tech industry and her accomplishments range from writing software for telephone switches to launching a new global satellite communication service. Anu has extensive experience in strategic marketing, corporate communications, demand generation, content marketing, product management, product marketing and technology development.