Knowing your customers – the key to combating fraud

Machine learning could be a powerful tool to combat fraud.

Today’s fraudsters are more sophisticated than ever, making catching them more difficult than ever. Increased access to data gives them the opportunity to attack online and offline in many new ways. 

In order to have the ability to fight the knowledge, resource and dexterity of modern fraudsters, developing an expanded view of your customers is critical. Effective management of fraud and corruption risks should focus on the effective detection, response and prevention of incidents. The unfortunate reality is that perpetrators of these crimes are constantly adapting their approach and as such the risk profiles of organisations require constant monitoring and assessment. Collaboration between the product development, marketing, fraud teams and third parties is essential, a collaboration driven by the aim of growing and protecting business in a sustainable way. 

To reduce fraud losses, companies turn to increasing security measures, which bring with them a certain amount of customer disruption. One unfortunate by-product of this is that in their endeavours to stop fraudsters, businesses can run the risk of treating good, honest customers like fraudsters. The biggest concern here, obviously, is the resulting irritated customers. Modern consumers have clear-cut expectations, they expect engagement with a brand to be easy and transactions they make to be safe. The fight against fraud can create friction in this experience, but it needn’t be that way.   

360 degree consumer view 

A multi-layered approach to authentication is regarded as the gold standard for identifying legitimate customers. The challenge here is striking the right balance of questioning so as to prevent an adverse reaction from the customer. Having access and insight into universal consumer behaviour down to the transaction level is necessary for fraud mitigation in the future and builds the framework to create a positive experience for consumers.   

The importance of profiling a customer runs deeper than only transactions and purchase patterns. It’s important to also consider their overall patterns of interaction including their preferred devices, items removed from their online shopping carts, how much time they spent on a site, and beyond.   

These patterns help merchants distinguish trustworthy activity from that which is not, a critical factor for determining the overall level of risk or trust that the merchant should have with that customer when building their profile. 

One of the keys to protecting yourself and your business is to take time to assess both the customer’s profile and what they are requesting. Although clichéd, there is a lot of sense in remembering that if it seems too good to be true, it probably is. 

Expanding profiles with a unified ecosystem 

But to gain true knowledge of a customer, businesses need to know what they’re doing both on and offline, not only with their business but others that individual engages with. This is where a unified ecosystem comes into play. It’s a vital component of customer profiling and many businesses now aim to achieve a single customer view by collaborating across internal silos to bring together information.   

However, a company’s single-customer view may still achieve only a partial view of the consumer given their view is constructed solely based on their relationship, rather than the consumer’s relationship with other companies. This is where the fraudster has the advantage. They have a much broader view, and businesses need to match this. Participating in a blended ecosystem by working with vendors, customers, partners and even competitors can bridge disparate data and internal silos, providing an enhanced will customer experience that supports business growth, without sacrificing protection. 

Finding the fraud solution that fits  

The growth of mobile and web traffic is increasing the volume of online purchases and, paired with the introduction of EMV, means more fraud is being pushed there. [EMV refers to chip cards whose improved security at physical stores may drive criminals to target online retailers.] It’s also making it more difficult to recognise your true customer. The rate of disrupted legitimate traffic to actual fraud attempts is at a staggering high of 30 to 1 – putting too many legitimate customers at risk of being challenged to make one catch. 

Businesses need to apply fraud mitigation strategies that reflect the value and level of confidence needed for each transaction to strike the balance of keeping customer disruption down while maintaining the necessary level of fraud management. It’s about finding the right size fraud solution.   

Machine learning - a powerful predictor? 

While technology is instrumental in increasing opportunities to commit fraud, the good news is that it can also play a major role in developing new methods and strategies that can be used to detect and prevent frauds. 

With more data comes more opportunity. As more people turn to digital solutions for everyday activities, it will generate huge amounts of data that forward-thinking businesses can use to identify trends and highlight suspicious behaviour. As a result, machine learning has become an invaluable advanced tool in the fight against fraud. While this is a great technology, a solid machine learning-based solution requires specialised expertise to apply rigorous methodology in data analysis. The systems begin to over-process every transaction that comes through, creating hundreds, even thousands, of rules that need to be managed. Rather, businesses should take a hybrid approach where tempered risk thinking can be applied to all of that great artificial intelligence. 

Balancing machine learning techniques with using characteristic-based analytics can help to reduce false positives and unhappy customers. By comparing current transactions with past behaviour, businesses are able to predict what tactics criminals are using before losses become apparent, and thus gain the upper hand in the battle against fraud.   

When it comes to fraud, it’s essential to always think ahead. Fraud detection analytics need to be able to learn and adapt quickly. Businesses do this when it comes to revenue growth and marketing to new customers, and fraud detection and prevention are no different.   

Boris Huard, Experian 

Image Credit: Gustavo Frazao / Shutterstock