Do You Still Need Manual Fraud Reviews When You Have AI?
With artificial intelligence and machine learning driving automated fraud prevention technology, are manual reviews dead? The answer is more nuanced than yes or no. In this article, we take a look at manual fraud reviews, what makes them valuable today and possible future scenarios to consider in planning your modern fraud strategy.
What are Manual Reviews?
Manual fraud review is a fraud prevention approach that relies on humans experienced in online fraud investigation to evaluate incoming transactions and determine a course of action, typically whether a transaction should be approved or declined. In the early days of online fraud prevention, most, if not all, transactions were manually reviewed, which consumed time and resources and was difficult to scale.
Automated fraud prevention screens and determines the vast majority of transaction decisions. The danger in relying solely on automated fraud detection is the potential for false positives, which means legitimate customers can be declined. This damages brand reputation and turns customers away — sometimes for good. Meanwhile, fraudsters can adapt their tactics and fool automated fraud technology when there is no human oversight of outliers that may exist on an order.
Today, what makes manual fraud reviews valuable is the ability to adapt and apply data from years of experience into a decision. As AI technology progresses, it may eventually be capable of making all transaction decisions, but today we still need humans to step in and provide wisdom and understanding of the full context. The bottom line is that manual fraud reviews are a necessary component of eCommerce transactions and need to be part of a multi-layered fraud prevention approach.
The Manual Fraud Review Process
A manual fraud review is typically triggered when the system detects a transaction as being potentially fraudulent. Machine learning algorithms have evaluated the details and scored it based on risk but cannot determine a final disposition because the information is unclear. This will result in an order being flagged for manual review.
Once the manual review occurs, the transaction goes to an employee, often a fraud analyst or data specialist, that can access the transaction and evaluate all the known data. They then determine whether to approve or deny the transaction. It is a time-consuming, involved process that includes:
- Evaluating incoming data and checking for accuracy
- Comparing the data to historical records and across the broader network
- Verifying details, performing link analysis, looking into relevant data such as phone number, and credit card information
- Communicating with the customer or bank if needed
- Investigating previous claims or issues from previous decisions
- Making a final decision based on all available data and an understanding of the customer behavior
Every case is unique and may have one or many reasons for why it was flagged for manual review.
Manual Fraud Review is Important to Customer Experience Outcomes
It’s easy to assume that because AI fraud detection can scale and perform faster than humans that automated fraud prevention is enough. Automated processes relating to fraud detection is invaluable and essential to fraud management today, but we still need humans to investigate and resolve more complex situations.
Without manual reviews, eCommerce retailers will often have high false decline rates; this insults good customers who may choose to do business with a competitor. A high insult rate means your brand is turning away legitimate customers and losing their business. If you have a high insult rate and are unaware of it, you are losing a significant amount of business without even knowing it.
A manual review process can prevent this.
Benefits of a Manual Fraud Review Process
There are multiple benefits to having a manual review process in place:
- Better Customer Experience. When a customer’s transaction requires evaluation, fraud expertise is a must. Conducting a thorough evaluation, often times without ever contacting the customer, is critical in providing a seamless experience.
- Improved Data Health. Machine learning and AI algorithms rely on the data they learn from to process future transactions. If they are fed inaccurate or bad data, they will generate higher false negatives and false positives. Left unchecked, this process can run rampant without the retailer knowing, leading to unintended loss of good orders, revenue, and customers.
- Lower Insult Rate. Again, lowering the insult rate means more legitimate customers are getting approved which improves revenue and lowers customer turnover.
Challenges of Manual Fraud Review
Clearly, AI-based fraud detection delivers speed, automation, real-time processing, and scalability advantages. Manual fraud review is not designed to compete with those capabilities. It is designed to complement AI tools to leverage the benefits of human wisdom, empathy, and judgment. However, there are challenges with having a manual fraud review process that need to be addressed:
- Requires Specialized Human Resources. You need to recruit and hire qualified fraud analysts with experience in data analysis, payment processing, fraud detection, and good communication and empathetic skills. It can be difficult to find qualified analysts, though remote work has opened up the job pool to wider geographies. Qualified analysts require commensurate compensation, which adds to labor costs.
- Is a Time-Sensitive Process. Just like the chargeback process, manual fraud reviews are time-sensitive. They depend on analysts being able to make decisions fast enough to keep customers satisfied. Impatient good customers may end up requesting order cancellations, which only complicates matters and compounds costs. Fraud analysts need to have a process that enables them to quickly gather the information needed, understand the situation and risk level, and make a decision.
- Can Be Difficult to Scale. When peak season hits, it can be exceedingly challenging for fraud teams to scale fast enough to meet high transaction volumes. This can lead to rushed decisions that may not be accurate and increased fraud losses.
- May Be Budgetary Constraints. Managing an internal manual review team is costly, particularly if you need to staff up to meet peak season demands.
When to Outsource Manual Fraud Review
Given the challenges of running an efficient, fast, well-trained manual fraud review team, many eCommerce retailers choose to outsource their manual fraud review process.
Radial offers a managed fraud prevention solution, consisting of a data scientist and an advanced analytics team. This team of experts complements the manual review team. The result is a very low manual review rate and high conversion rates. We have the capacity to scale and meet an eCommerce retailer’s needs.
Radial provides manual fraud review, fraud chargeback resolution and indemnification, zero fraud liability, 70+ dedicated employees, and 20+ years of expertise.
Follow Radial on LinkedIn, Facebook and Twitter.
Want to hear more about Radial’s fraud prevention solutions?
