Don’t end up by only having fuzzy dots…

One of the main effects of the Fourth Industrial Revolution is increased human productivity and creativity. Over the past year we have seen companies continue to be exposed to massive fraud, corruption and cyber security attacks. To continuously outsmart the fraudsters, who keep on perfecting their tactics due to new emerging technology, organisations need new, efficient and effective tools such as visualized big data analytics. Although data is available in abundance and it provide enormous potential to deploy analytics and uncover what mechanisms fraudsters use, fraud analytical models should satisfy various requirements. The amount of data produced by any large organisation is growing at a rate of 40% to 60% per year. Simply storing this huge amount of data is not going to be useful if it is not used smartly.

By using data smartly, using intelligence analysis to combine datasets and detect individual’s relationships, connections and time-lines using visualization tools the initial fuzzy dots can respond to the data evolution by means of cost-effective and efficient tools, multidisciplinary insight, re-engineered systems and processes, response and prevention (data-to-decision process) of fraud. By layering data modelling onto various forms of customer data, employee data, exception reporting, financial trends, geospatial data, service provider data, historic fraud data, etc, it becomes possible to develop new fraud prevention strategies based on machine learning models. To make sure that the models continuously performs at its best, they should be fed with the most accurate and relevant data available.

Many organisations still believe that customer behaviour is independent and that internal fraud activities are non-existent. Fraud is a social phenomenon and fraudsters often, collaborate, associate or bond with similar others. Hence, to fully exploit available data by also defining networks between various data platforms.

Payroll fraud, particularly in large companies where well-intended controls exist, is difficult to detect. They are often challenged with large numbers of employees (full and part-time), contractors, multiple offices, with often legacy and different platforms used for payroll processing and multiple payrolls running every month.

XTND was approached by a client in Gauteng who supplies cleaning services to their clients. They have between 3000 – 3200 employees of which 40% are employed on a permanent basis. The rest were temporarily employed. Data from the wage’s payroll over a period of 20 months were transformed into one master tabular source to be imported into IBM i2 Analyst’s Notebook. An import specification for the data file was created and entities (employee names and bank accounts) created links. More information was added to the entities to include employee numbers and amounts paid which allowed for a more positive identification of the employee and quantifying the possible loss.

Algorithms combined various datasets and it was established that temporary employees who moved over to permanent employment were never taken off the temporary staff payroll system and therefore two salaries were paid under two different employee numbers and two different bank account numbers used. The bank account number on the contract for the temporary staff member was simply changed by the financial administrator upon offer for permanent employment was accepted. The business lost R890 000 on salaries alone for the 20-month period.

The data further revealed a network between the financial administrator and three service providers and upon further investigation it was confirmed that fictitious invoices were issued. The business lost a further R630 000.

Networks are important to disentangle complex fraud patterns and those who are willing to commit fraud do not discriminate. Payroll fraud can result in huge financial losses, reputational risk and the deterioration of ethical behavior amongst employees. The cost to identify and prevent fraud is less expensive than the cost of the fraud that gets committed. Connect the #fuzzy dots, get the picture and be in control.