By : Dena Hamilton
As the rate of fraud continues to increase, being able to detect such transactions and stop them before they are completed clearly needs to be a top priority for banking institutions.
With the overall number of transactions rising hugely, and developments such as real-time payments helping make settlements faster, the solutions banks have in place for fraud detection are coming under more pressure than ever. In many cases, these systems will only have a matter of milliseconds to determine whether a transaction is genuine.
The good news for banks, however, is that technological advancements can provide them with many more options for meeting these challenges, thanks to a new generation of big data analytics and machine learning applications.
More data, more opportunities
As more people turn to digital solutions for all their everyday activities, including banking and making payments, they will generate huge amounts of data that forward-thinking banks can use to identify trends and highlight suspicious behavior.
Algorithms are able to look at a wide variety of metrics to evaluate a transaction, from the trustworthiness of a vendor, through to whether the time and location of the activities matches an individual's previous behavior. And the more data points there are available, the more accurate the decision-making.
Why machine learning helps
With millions of transactions taking place every day, it will be impossible for humans to study what is going on and identify patterns. Therefore, the analytics tools themselves must be able to evolve, taking into account new information and comparing it to past data in order to make decisions about the validity of transactions.
The impact of this can be clear. It was recently noted by CIO.com.au that PayPal, for instance, uses machine learning technology that studies users' purchase history. Once it spots patterns, it can implement new rules that prevent scams being repeated.
The result of this is that PayPal has a revenue fraud rate of just 0.32 per cent compared with the industry average of 1.32 per cent.
This level of analysis, studying the buying behavior of individual customers, means that the system understands the type of items that users normally buy and can flag up as suspicious those that deviate from this pattern. Obviously, this alone is not a guaranteed indicator of fraud, as there can be many reasons why an individual would make an unusual purchase, but by combining this with other indicators, it can build up a complete picture of a transaction.
Keeping up with the fraudsters
Being able to adapt and evolve strategies is vital in keeping up with the tactics used by fraudsters, as these will change continually in order to evade detection. For example, if a system is put in place to flag up any payments over a certain amount for a more in-depth review, criminals will quickly learn to place transactions that come in just under this limit.
Tony Ippolito, the strategic risk and technology manager for eBay Enterprise, explained: "When you close off a certain area, you have to be aware of what the next logical step for them is. If you shut down overnight shipping, then they'll move into third-day shipment. It's a lot more nuanced than that, but that’s the general idea."
This is why fraud detection analytics have to be able to learn and adapt quickly. By comparing current transactions with past behavior, they can help predict what tactics criminals are using before losses become apparent and gain the upper hand in the battle against fraud.