Fraud is a growing concern: the global market is flooded with a variety of specialised fraud solutions i.e. solutions that focus on one specific type of fraud, which means fintechs, FIs, and merchants have to evaluate the decisions and alerts coming out of each of these solutions in a siloed and non-contextualised manner. Without having a risk engine or a platform hub that can harness the decisions seamlessly across these various detection points, it can either lead to ineffective decisions or more friction with customers.
The cost of vetting and partnering with individual vendors and fitting them within the fraud operations ecosystem is rising, and at a time when budgets are limited, filling the gaps between fraud solutions with manual operations only ends up becoming more expensive.
Siloed and out-dated fraud strategies do not scale
Many of the fraud strategies are focused more heavily on the initial verification of new accounts or lean too heavily on the monetisation points like payment transactions. Leaving the key entry points across the account lifecycle unguarded usually leaves the fintechs, FIs, and merchants vulnerable to fraud attacks, as well as possible fines, not to mention the reputational damage.
Account and transaction fraud are interconnected and should therefore be addressed holistically. The information available at the account login is valuable when making a decision on the payment authorisation. Behavioural biometrics are useful inputs for making transaction decisions, as these can be used to establish the reputation of the device and email address. As most fraud teams will tell you, they want more data, not less, when making the decision to approve or decline a transaction.
Many fraudsters take their time to perpetuate the fraud: they wait and watch the account, trying to understand the patterns on the account and mimic the behaviours to avoid suspicion. Rules-based fraud strategies do not scale for large-scale pattern analysis to detect microscopic anomalies and hidden linkages.
With more and more of the payments going real-time and money getting settled instantly, there is a growing need to detect bad behaviours even before payment happens. Understanding more of the account behaviour upfront and with speed can help businesses predict patterns of transaction fraud and prevent it.
Emergence of orchestration hubs
Orchestration hub is more than a buzzword and it is becoming more of a reality recently with the exponential surge in fraud. Hub solutions allow for the seamless integration and orchestration of detection inputs, third-party data, and data consortiums via a sophisticated machine learning pipeline.
The below table shows some of the pros and cons when considering a hub solution:
It is crucial for fintechs, FIs, and merchants to follow best practices and keep themselves one step ahead of bad actors. Here are some key strategies that help proactively detect fraud:
Anomaly detection: Changes in account usage patterns is a key indicator of fraud. It is important to look for obvious irregularities in account activities which may suggest that a user is exploitative. These may include transactions which are done during non-regular hours, or orders placed with higher quantities than usual for a specific product, specific account changes, using a different device, or accessing the account from proxy IP etc.
Digital footprint assessment: There are 4 pillars of the digital footprint: device fingerprint, IP address, phone, and email. These can provide crucial signals to understand the origins of the login or payment and make a decision on authorisation.
Data-driven machine learning strategies: Models that use features and profiles and target user behaviours, session information, order history, and transaction data as key levers are far more effective than rules-driven reactive strategies. These detect linkages between account and transaction attributes like IP, card numbers, device used, same recipient, shipping address etc. with machine learning algorithms and techniques.
Decisioning speed: Real-time scale and speed to go through millions of data points in a matter of milliseconds and provide a decision with low false-positives is key to minimise liability. It is also important to be able to refresh your models on a frequent basis to be on top of new patterns of fraud.
Checklist to consider when selecting a fraud solution:
Does it have proven KPIs with existing customers? Can it support your business outcomes?
Does it have a broad coverage of attack vectors with a multi-layered orchestrated approach that protects most entry points starting from account to payments fraud?
How quickly can the machine learning models adapt to new fraud trends and how soon they can be deployed into production?
Do behaviour data collection, deep link analysis, biometric authentication, document verification, liveness checks, machine learning assessment, rules all work seamlessly with each other?
Does it scale well to volumes and is it fast?
Does it have a stake in your success and growth?
Remember, the fix shouldn’t cost more than the problem.
Reach out to us for more information on how we can help protect and grow your business.
Srividya Sunderamurthy
srividya.sunderamurthy@trustvesta.com
About Srividya Sunderamurthy
Srividya Sunderamurthy heads product strategy at Vesta Payment Solutions. She has over 20 years’ experience leading end-to-end product strategy and building, launching fraud, AML solutions for banks, fintechs, and ecommerce.
About Vesta
Vesta is a fintech pioneer in fraud protection and fully guaranteed payment technologies, helping online merchants, major telcos, payment processors, and acquirers optimise revenue by eliminating the fear of fraud. The company’s flexible, scalable solutions enable companies to grow their businesses by focusing on revenue rather than risk, delivering secure, frictionless transactions that maximise acceptance and improve customer experience – all backed by a zero-fraud-liability guarantee.
Every day we send out a free e-mail with the most important headlines of the last 24 hours.
Subscribe now