Using analytics successfully to detect fraud
KPMG’s Data Analytics Lead in South Africa, Frank Rizzo, predicts that using anti-fraud technology will become increasingly important. Integral to the success of this technology is trusted analytics.
Fraud-focused analytics is a powerful tool that can assist companies in detecting potential or on-going fraud. However, a KPMG survey found that while technology enabled 24 percent of fraud cases, technology was used to detect fraud in only 3 percent of cases.
Why aren’t more companies employing analytics successfully to catch fraudsters?
When one considers the large amount of data that a company produces, it would seem logical that more would also be using analytics to make sense of it.
Frank Rizzo, D&A Lead for KPMG in South Africa, says one of the reasons behind companies’ hesitation to use anti-fraud technology is due to a lack of the necessary skills and know-how.
“I think the issue is around ‘what data do we use?’ and ‘how do we do this?’ To me, an opportunity exists to collaborate across industries in South Africa so that we can leverage the limited skills in this space,” says Frank.
Another reason for the lack of adoption of anti-fraud technology is a lack of trust in the effectiveness of anti-fraud programmes.
“Like any data analytics solution, there is a huge dependency on the quality and richness of the data available. If this is not high enough, you land up with ‘false positives’ and a solution that hampers a business process as opposed to helping manage risk,” says Frank.
Many companies do not trust the ability of analytics to distinguish between legitimate transactions and fraudulent activity. To encourage better adoption of analytics in the fight against fraud, this trust relationship needs to be strengthened.
Trusted analytics is integral to ensure effective anti-fraud analytics
Creating a trusted analytics programme to monitor and detect fraud is becoming more important as technology continues to play a bigger role in fraud. It is becoming so important that the benefits of employing anti-fraud technology far outweigh the potential risk, says Frank.
“Even thieves are moving to the digital world! The cost/benefit of any solution needs to be analysed and is dependent on how much money a company is currently losing to fraud.”
The four anchors of trusted analytics
To ensure the success of a trusted analytics programme, Frank suggests considering the four anchors of trust.
Data analytics is a complicated process and its accuracy relies on the quality of the data. This means that data must be relevant and demonstrate a business understanding of the particular industry. Data must also be accurate and up-to-date in order to distinguish “normal” data from anomalies.
Ability to learn
A successful analytics programme walks a fine line between generating too many and too few red flags. Care must be taken in refining the algorithm to achieve this balance. The learning potential turns a good solution into a great solution.
Regularly updated programmes
Analytics programmes have to be updated regularly as circumstances change in order to avoid generating too many false positives, which take time and resources to investigate.
Buy in from employees and business partners
The most important of the four anchors concerns getting everyone on board. Companies need to convince their employees and business partners that anti-fraud detection is necessary. Frank suggests making this buy-in into anti-fraud technology a condition of employment – something that is happening more and more in high-value, high volume environments, e.g. a bank’s trading desk.
“Externally this becomes more difficult to achieve, but I believe this is a differentiator for businesses. They need to get the message out that they don’t do fraud, they have a number of anti-fraud systems in place and if you want to do business with them, you need to comply.”