Few companies succeed in employing systematic data analysis to detect fraud. What’s necessary to build trust in an analytics based approach to fraud detection? KPMG’s report “Using analytics successfully to detect fraud” explains the challenges of managing an analytics-driven program and examines the steps companies should take to improve the chances of delivering such benefits as protecting the reputation of the organization.
Why the low usage of analytics?
In our recent report the key finding in relation to Data Analytics is that only 3 percent of the cases were detected using proactive, fraud-focused analytics. In the era of digitization this seems to be going against the trend of deploying more and more analytics in many business processes. If an analytics-driven anti-fraud program does not successfully detect cases of wrongdoing in the early phases, management’s confidence in analytics as a valuable tool to pinpoint fraudulent activity could well erode. Let`s explore a little the reasons that go behind this low number of detection using analytics.
Some corporate decisions makers do not understand what analytics can do for them. Other balk at the expense. Still others may believe that until a major fraud occurs at their company, it is not worth investing in advanced analytics to detect their potential wrongdoing before it occurs.
I believe that this lack of adoption also reflects a ‘trust deficit’ – a lack of trust and confidence that the underlying data, the analysis and the business interpretation of the outcomes will be able to distinguish between legitimate transactions and fraudulent activity in an efficient and cost-effective manner.
Trust dimensions and anchors
- Successful analytics requires high-quality components: The first trust anchor relates to the quality of the components in the analytics program. The data has to be accurate and up-to-date. The sources of the data need to be known and understood.
- Knowing what is normal: A successful fraud detection program through analytics must consider detecting both anomalies and knowing what is normal. When analytics-based fraud detection programs fail, it is often not because they lack analytical rigor but because the implementation platform lacks the knowledge of what is expected to be normal.
- False positives must be carefully managed: Is the output accurate and useful in the sense of fulfilling its purpose? A successful anti-fraud analytics process has to walk a fine line between generating too many and too few red flags. Refining the algorithm to achieve this balance is a process of trial and error.
- Operational control must be sustainable: For an analytics program to be effective, it is not sufficient merely to design an algorithm and then leave it untouched to operate indefinitely. Rather, it has to be updated regularly as circumstances change.
- Anti-fraud analytics must be ethical: Is its use considered acceptable by such stakeholders as employees, suppliers, customers, business partners and regulators? This, we believe, is the most important of the four anchors because it addresses some of the most sensitive areas of the relationship between the company and its stakeholders, in which trust plays a vital role.
- Building a better culture: Societies and the companies within them are experiencing a trend toward greater transparency. Stakeholders are demanding more openness from companies and other institutions. People must be confident that the analytics algorithms work as intended and must trust each other to use them properly.
Successful fraud detection using analytics requires high quality components, effective use of analyzing transactions, long-term operational control and ethical integrity of the process.
Actions points to consider
To get trust in the use of analytics to detect fraud it is important that the rationale of applying analytics is being communicated clearly. Inform your stakeholders what you are trying to protect (reputation, monetary value, trust with 3rd parties, etc.) and that applying analytics is a way to prevent reviewing every transaction manually and in detail. The normal business transactions will not be an anomaly so the effective use of analytics allows a company to focus only on transactions that are unusual. If this message is communicated to all stakeholders one will have buy in and acceptance for the use of analytics.