Analytics detect fraud before it happens

Companies that extend their budgets to allow for increasing levels of fraud are doing nothing to effectively or sustainably combat the problem. The solution is to use statistical modelling to detect the likelihood of fraud and prevent it.

By Colin Hill, senior solution manager: governance, risk and compliance and financial crimes at SAS Institute.

Fraud has a massive financial impact on South African companies, and the retail sector is hit especially hard. Despite this, companies are not being proactive enough to protect themselves, their profit margins and their reputations.

Most companies budget for the costs of undetected fraudulent transactions, and measure their success in combating fraud against these budgets. Unfortunately, fraud levels are increasing every year, and repeatedly extending the fraud budget is doing nothing to prevent these activities.

Companies need to acknowledge that they are being defrauded, and they need to ensure that they have fraud detection tools in place. These tools should take the form of statistical models to proactively identify and prevent fraud at transaction level. In other words, they need to detect potential future fraud by learning from the fraud that has occurred in the past.

The applications of scoring fraud

In Europe, companies are using statistical methods to develop a profile of the characteristics of the typical fraudster to then generate a fraud score of any individual. For example, if a person applies for a loan and a bank finds that they reach a specific number on the fraud score, the loan might not be granted, additional due diligence could be done or the account could be flagged to monitor for signs of fraudulent activity.

However, the fraud score is not specifically designed to discriminate against individuals, but rather, to look for patterns. This is already happening with banks monitoring credit or debit cards and flagging multiple transactions in fuel, clothes or food outlets in a limited space of time. So, for example, the bank could phone a customer if their card has been used at a filling station several times within an hour and will be able to prevent further transactions from taking place.

Companies could also use these fraud scores in the recruitment process. Most employers in South Africa do a credit check on potential candidates - but this is not linked to criminal or fraud characteristics. If they have an accurate view of what a typical fraudster looks like, they stand a better chance of avoiding appointing risky employees.

Beyond forensics

The forensic departments of many companies are made up of ex-policemen. While they are brilliant at interviewing, assessing and catching criminals, they have little or no experience with rule-based or analytical detection engines. This places them at a disadvantage because they are not able to correlate historic fraud data and develop analytical models to proactively combat fraud in an organisation. A lot of industry education is still needed to highlight the value of analytical modelling capabilities to organisations.

Once the practical value of these kinds of capacities have been realised, organisations will begin to budget for them, instead of simply increasing their budgets to absorb the rising fraud levels.

Protect ourselves better

Of course all the fraud detection capabilities in the world mean nothing if individuals do not learn how to protect their personal information better. The banks can do all they can to prevent fraud, but if people do not take responsibility for their own information, all the banks' efforts will be futile.

Individuals need to start managing and protecting themselves better in collaboration with institutions, banks and retailers if we are to have any hope of proactively combating those who want to defraud us.

  • bullet
  • bullet

subhead header 3



About SAS

SAS is the leader in analytics. Through innovative analytics, business intelligence and data management software and services, SAS helps customers at more than 80,000 sites make better decisions faster. Since 1976, SAS has been giving customers around the world THE POWER TO KNOW®.


Editorial contacts:

Back to Top