James Loft, Chief Commercial Officer, Rainbird
Businesses are struggling to keep pace with relentless and ever-shifting fraud attacks. In the last year, the number of reported fraud cases nearly doubled. The ISMG reported that only 34% of C-level leaders have high confidence in their organisation’s ability to detect and prevent fraud.
Worryingly, most existing fraud detection tools are not up to the task of competing with this malicious behaviour and businesses are finding it difficult to strike the balance between security, customer service expectations, and regulatory requirements.
Despite this, demand for omnichannel convenience – where consumers want the ability to shop anywhere, anytime, regardless of their preferred means – shows no signs of respite. According to Invoca 75% of consumers say it’s important or extremely important to be able to switch easily between channels when interacting with their bank.
But organisations must ensure they don’t become preoccupied with providing omnichannel solutions and forget about fraud prevention in the process. The more channels, the more siloed data there is to track and the more potential for points of vulnerability in the security of these.
All the while, the velocity of real-time payments and regulations such as the Second Payments Services Directive (PSD2) put another level of pressure on organisations with a higher data throughput. This disparate view and data on a customer and a potential fraud case can also mean that investigators and analysts have multiple views of customer risk. In turn, this can affect the time it takes to review cases across different departments and impact legitimate customer journeys.
Additionally, financial institutions also need to ensure they aren’t introducing any friction into fraud processes, otherwise this might impact customer retention. Especially as nearly 40% of cardholders abandon cards after false declines, according to Javelin, and a quarter of these people move their cards to the back of their wallets.
Often when transactions are falsely deemed to be fraudulent this is because of outdated tools that can’t tell the difference.In order to mitigate this, organisations need a single view of the customer journey and potential fraud access points, uniting organisational silos to deal with the different stages of fraud.
To solve this, many organisations are adopting machine learning tools, which are effective at detection but aren’t capable of investigating potential fraud cases. To make a nuanced judgement on how best to resolve a fraud case, you not only need the data-up nature of machine learning, but also the human-down element. This turns the process from a task automator into a decision automator.
One such option is to house fraud processing under an automated decision-making platform that is built on human expertise and company best practice and can seamlessly integrate with multiple interfaces and solutions. This takes the experience of a fraud professional and applies that to thousands of cases with the use of artificial intelligence (AI) at scale.
As a result, data and logic is set by business experts, rather than the huge, hard-to-fathom datasets of machine learning.It also means that each fraud decision is transparent and accountable and that the end decision is still made by a human after looking at an audit trail of automated decisions.
A company’s fraud expertise is then encoded in this single fraud decisioning platform, meaning that the highest standard of fraud processing can be applied across various channels and large data sets to deal with multiple cases simultaneously. Additionally, it provides a single view of the customer journey and potential fraud access points, delivering a holistic approach to detection and investigation.
Consequently, cases are more likely to be correctly spotted, less likely to be falsely flagged, and can be dealt with faster. The end result: fraud detection is more subtle and adaptable, efficiency is increased, there’s less friction in transactions, and customer loyalty is improved.